This isn’t to say that loyalty programmes aren’t an immensely important and profitable part of the airlines business, but it’s become a hit-and-run critique of specific airlines and the industry as a whole. We wanted to tackle this question in a sober manner by outlining the financial accounting practices of airline loyalty programmes, not just to dispel the myth, but also to allow readers to better appreciate the actual importance of loyalty programmes.

Read more

Global air travel has been significantly impacted by the outbreak of the Middle East conflict on 28 February 2026. The closure of airspace in multiple countries led to the temporary grounding of several Middle Eastern airlines and disruptions to flights that transited through affected airspace.

While the affected airlines have attempted to mitigate disruptions, the impact continues to be felt a month later. During some previous analyses over the last month we’d framed mitigation strategies around disruptions “persisting”, and it’s fair to say that things are certainly persisting!

A unique challenge for Australia is the significant impact that these disruptions have had, and continue to have on travel between Australia and Europe. This is ultimately due to the dominance of Emirates, Etihad Airways and Qatar Airways (ME3) on these routes. We’ve previously highlighted how the ME3 accounted for more than 50% of the total Australia-Europe market in recent years.

With the ME3 running a skeleton operation with very little connecting traffic passing through their hubs, and passengers showing significant risk aversion to flying through Abu Dhabi, Doha and Dubai, there’s a significant capacity crunch occurring. Australia-Europe isn’t the only market affected, with the same constraints affecting most east-west traffic between Europe and Asia.

Why is Australia-Europe so important?

According to ABS data, 3.6 million people traveled between Australia and Europe in 2025, with approximately 44% traveling between the Australia and the UK and Ireland. The remaining 56% were fragmented across Europe, with Italy and France the next largest markets accounting 11% and 9%, respectively.

The market is marginally dominated by outward bound travel, with 54% comprising Australian residents and 46% foreign visitors. This varies significantly when considering the UK and Ireland separately to the rest of Europe. Outward bound travel accounts for 46% of the UK and Ireland market, whereas Australians account for 61% of travel between Australia and the rest of Europe.

The distribution/fragmentation of the outbound market also varies, with the UK and Ireland only accounting for 37% of travel by Australian residents. Italy and France comprise 16% and 9%, whereas Greece is a significant 8% of the market.

This has turned Europe-Asia into the bottleneck in Australia-Europe travel. There is ample capacity available between Australia and Asia, but simply no seats available between Asia and Europe. For example, there are seats available on nearly all of Cathay Pacific and Qantas’s flights between Australia and Hong Kong over the next week (Friday 27 March though Thursday 2 April), but not a single seat available on any of Cathay and British Airways’s flights between Hong Kong and London over the same period (other than the odd spattering of full fare first tickets). Similar patterns are found through other Asian hubs to most European cities.

Impact on Qantas

Qantas have already run into several challenges. Firstly, they move a significant amount of European traffic through their Emirates joint venture (it’s not just a codeshare). Ultimately, passengers that they are ticketing are their responsibility and are having to find alternative routings to get them to their destination.

They have limited redundancy capacity to/from Europe, with daily flights from Perth and Sydney to London, 3x/week Perth-Paris flights, and 3-4x/week seasonal Perth-Rome flights that are only due to recommence from the start of May.

Qantas’s European flights:

QF9/10 Perth-London Heathrow (Daily) non-stop B787-9
QF1/2 Sydney-Singapore-London Heathrow (Daily) A380
QF33/34 Sydney-Perth-Paris (3x/week) non-stop B787-9
QF5/6 Sydney-Perth-Rome (3-4x/week; seasonal) non-stop B787-9

As noted, redundancy capacity through Asia is also limited given the significant capacity bottleneck it now represents. It’s been widely reported that a lot of traffic is now routing through the US, despite the practical challenges as international transit passengers are required to clear US immigration.

Qantas have also had to reroute its non-stop Perth-London flight via Singapore as airspace closures have increased flying time and compounding payload restrictions. Without the fuel stop in Singapore on the westbound leg, the additional payload restrictions would limit passenger uplift significantly. We covered this is a detailed post a few days ago.

Qantas have been on record to indicate that they are considering options to add extra capacity to Europe, so we thought we’d look at the options they might consider. There are lots of rumours going around what they might do and while some of these link to the options we look at, the purpose of this post isn’t to test the veracity of them, rather to consider the relative challenges they face.

Option 1: Increased frequency on Perth-Paris/Rome

The first, and possibly simplest option would be to increase frequency on existing Perth-Europe flights. At present, Perth-Paris operates 3x/week year round with on a B787-9 aircraft, originating in Sydney (QF33/34), while Perth-Rome operates seasonally 3-4x/week, also originating in Sydney (QF5/6) and operated with a B787-9 aircraft. Notably, the Rome flight will run a longer season this year, also with an earlier start date, significantly increasing capacity over last year. This increase was already announced last year, and isn’t a reaction to current events.

While additional flights could be added on an ad hoc basis without significant network changes elsewhere, sustained increases in frequency will require the B787-9 to be taken off other services. This would either require reduced frequencies on other routes, or switching other B787-9 flights to other aircraft.

During the NS2026 season, other B787-9 flights include (excluding European flights):

• Brisbane-Los Angeles (daily)

• Melbourne-Dallas (3-4x/week)

• Melbourne-Los Angeles (5x/week with remaining 2x/week A380)

• Sydney-Vancouver (3-4x/week)

• Sydney-San Fransisco (3x/week)

• Sydney-Auckland-New York (5x/week)

• Sydney-Santiago (4x/week)

One option could be to switch BNE-LAX back to the A330-200 which operated the route after the COVID pandemic until last year, however this would be unpopular with customers. The A330-200 wouldn’t be viable on other routes as it would incur even larger payload restrictions than on BNE-LAX.

An alternative could be American Airlines taking over BNE-LAX to enable B787-9 reallocation to Europe. Notably, Qantas and AA did this recently with AA operating BNE-LAX 3x/week with B787-9s in December 2025 and January 2026 to enable Qantas to reduce frequency from daily to 4x/week and redeploying the aircraft to increase the frequency of MEL-DFW from 4x/week to daily during the same period. On the presumption that AA would have sufficient capacity this would be an elegant solution, although it’s a much easier ask during the southern hemisphere summer. The northern hemisphere summer is AA’s peak long haul season with a greater emphasis on deploying their own capacity to Europe.

Option 2: Add additional London flight via Asia

Another straight forward option would be to add another London flight, similar to adding more Paris or Rome flights. For example, a return of Melbourne-Singapore-London. This could be operate by the B787-9 or A380, requiring some additional aircraft time, however this would require less additional time than additional Perth-Europe flights as the Melbourne-Singapore (and return) sectors would presumably replace existing QF35/36/37/38 sectors. In doing so it has a somewhat smaller opportunity cost.

Routing through Singapore would present the best opportunities for network effects with existing connections, similar to the manner in which QF1/2 already operates. Qantas would have little problem with traffic rights at Singapore, while Heathrow slots could be temporarily sourced from any number of Gulf airlines who are currently not utilising their Heathrow slots (let’s just assume their joint venture partner Emirates).

Option 3: Switch Perth-London to A380

Another solution would be to increase capacity on the existing Perth-London flights by switching the B787-9 to the A380. At present, the outbound leg is currently flying via Singapore to enable higher payloads that are being undermined by longer routings due to unavailability of European airspace. The fuel stop in Singapore is somewhat enforced, and adds significant time to the flight. QF9 was renumbered QF209 to reflect the retiming of the Perth departure from 7:15pm to 3:50pm, maintaining the arrival at Heathrow at 5:05am.

QF9 PER-LHR 7:15pm 5:05am
QF10 LHR-PER 11:50am 12:30pm

QF209 PER-SIN-LHR 3:50pm 5:05am
QF10 LHR-PER 11:50am 12:30pm

The return QF10 remains unchanged, operating non-stop and with the same schedule. This generates downstream challenges with the earlier departure from Perth potentially undermining connecting traffic from the east coast, and limiting options to rotate aircraft to/from the eastcoast.

Switching from the B787-9 to the A380 would generate a large increase in capacity. Furthermore, there is somewhat more redundancy capacity in the A380 fleet compared to the B787-9 fleet. But there are some externalities, and a very big one: the return leg (QF10) wouldn’t likely be able to operate non-stop. At 16.5 hours it would require large payload restrictions. Comparatively, westbound QF8 DFW-SYD is Qantas’s longest A380 sector and is slightly shorter, typically coming in just above 16 hours, however this sector is notoriously payload restricted and QF10 would generate larger payload restrictions, nevermind the need to carry more fuel due to lack of nearby alternates for the A380 at Perth. As we see with the outbound leg, adding the fuel stop adds a lot of additional time (3.5 hours on the outbound leg). Taking the same departure from Heathrow would push the arrival at Perth from 11:45am to 3:15pm.

This would be catastrophic since the current combination of QF9/10 and even QF219/10 ultimately required two aircraft for a daily service. Switching to the A380 and adding a fuel stop in Singapore on the return leg would mean the aircraft arrives at Perth at 3:15pm with an outbound departure of 3:50pm meaning it would require 3 A380s to operate. This means that the B787-9s that might come off Perth-London wouldn’t be able to replace the A380s it displaces elsewhere on a one-for-one basis.

But something magical has happened. Just today we’ve seen that Qantas have retimed QF219’s Perth departure to 7:35pm with London arrival now 10:05am. The flight number has subsequently been changed again, this time to QF219. Once again, QF10 remains non-stop with an unchanged schedule. This change is effective from 29 March, coinciding with the start of the IATA NS2026 season. This has neccesitated a change in their Heathrow arrival slot, likely affected through a temporary slot swap with another airline or a temporary dispensation from the slot coordinator. Eiether way, this likely explains why the retiming of the Heathrow slot coincides with the start of the IATA NS2026 season.

QF9 PER-LHR 7:15pm 5:05am
QF10 LHR-PER 11:50am 12:30pm

QF209 PER-SIN-LHR 3:50pm 5:05am
QF10 LHR-PER 11:50am 12:30pm

QF219 PER-SIN-LHR 7:35pm 10:05am
QF10 LHR-PER 11:50am 12:30pm

The retiming reclaims ground time at LHR, allowing improvement of connectivity at Perth (or at least allowing the original connectivity to return). Furthermore, it gives Qantas better options to rotate aircraft to/from east coast.

But this opens the door to the A380. This is pure speculation: what if QF10 were to be replaced with the A380 and a fuel stop in Singapore introduced? Let’s call it QF210 and assume it adds 3.5 hours flight time.

This would put QF210 arriving in Perth at 4:00pm, allowing departure as QF219 at 7:35pm. While there are some connectivity issues related to the later departures, it would now require 2 A380s instead of 3. This would allow a straight swap between the B787-9 and the A380 on other routes (e.g. SYD-DFW/JNB/LAX/SIN)

Furthermore, aircraft rotation to/from the east coast is not a challenge as it could rotate with the QF1/2 aircraft at Heathrow, similar to how Qantas typically rotate B787s and A330s at other outstations (e.g. AKL, DFW, LAX, SIN).

So this is doable? Well, there’s one big catch. Qantas’s Perth terminal doesn’t have A380 capable gates and utilising the A380 would mean operate to/from T1, undermining connections and lounge options. Ironically, the emphasis that Qantas have had on engagements with Perth Airport regarding the importance of operating their international flights to/from the Qantas terminal would almost preclude this. But desperate times call for desperate measures?

Option 4: Extend Australia-North America flights to Europe

The most complex and even risky would be to extend one of existing North American flights onward to London. This has several challenges and options. At present, their North American network looks as follows:

BNE-LAX (daily)
MEL-DFW (3-4x/week)
MEL-LAX (daily)
SYD-AKL-JFK (5x/week)
SYD-DFW (daily)
SYD-HNL (5x/week)
SYD-LAX (daily)
SYD-SFO (3x/week)
SYD-YVR (3-4x/week)

Each option has its own idiosyncrasies. Firstly, we might remind readers that the US doesn’t have international airside connectivity meaning that connecting passengers much clear US immigration. That creates a range of externalities and even risks. In the past, some US airports had an international transit lounge meaning that same flight passengers didn’t have to clear immigration, however it seems as though this is no longer available. While passengers don’t have to collect bags and clear customs, passenger need to clear immigration and would require an ESTA or visa.

Secondly, it’s not operationally practical on all routes. We can exclude SYD-HNL since it would require a change from the A330 currently operating on the route, and also SYD-AKL-JFK since this would imply a 2-stop route to LHR.

To some extent, we might exclude SYD-SFO/YVR since its only operated 3x/week and wouldn’t generate any network effects without and BNE/MEL-SFO/YVR connections, making BNE/MEL-LHR a 2-stop itinerary. Another consideration for YVR would be the bilateral arrangement. The 1988 Australia-Canada air service agreement doesn’t make provision for points beyond Canada. However, this agreement has recently been updated, and while reported as an open skies agreement that enters into force at the end of March 2026, it’s not clear whether this is an open skies agreement that would allow points beyond Canada or just moving to unlimited frequencies (e.g. similar to Australia-Singapore that doesn’t have any capacity constraints, but still limits intermediate and points beyond).

That leaves us with DFW and LAX. Let’s presume the option is to extend SYD-DFW or SYD-LAX to LHR. Both operate daily with the A380, proving Qantas with a large capacity platform. LAX provides daily connecting feed from both BNE and MEL, while DFW only less than daily from MEL. Furthermore, the westbound DFW-SYD leg operates under significant payload restrictions, limiting the incremental capacity it might provide.

And finally, assuming LAX is the most practical option due to connectivity to/from BNE, MEL and SYD, a final challenge is the lack of schedule coordination between these flights. For example, if the SYD-LAX flight were to continue to London it would be too early for BNE connections, and far too late for MEL connections. This would require multiple pieces to be rescheduled.

QF11 SYD-LAX 5:50pm 2:30pm
QF15 BNE-LAX 8:35pm 4:30pm
QF93 MEL-LAX 9:45pm 7:05am

QF16 LAX-BNE 10:10pm 5:00am
QF94 LAX-MEL 9:50pm 6:45am
QF12 LAX-SYD 11:30pm 7:30am

Thus, we’re heavily discounting the US option due to the lack of connectivity, scheduling challenges and challenges with respect to international transit passengers needing to clear US immigration.

No need to speculate, here’s the answer

Correct as of 4pm Thursday 26 March:

Just as we were finishing this up we began to notice a number of schedule updates appearing, so no need speculate any further. These are schedule updates now showing and on sale!

Rome: Perth-Rome was due to recommence in early May with 3x/week flights, increasing to 4x/week from late June until late September. Instead, it will operate daily between 18 May and 26 July, with select flights not originating in Sydney.

Paris: Paris flights will increase from 3/x to 4x/week between 20 April and 17 May. Westbound routing will operate Sydney-Singapore-Paris, excluding Perth, although its uncertain whether this routing only applies to the additional flight or all flights. Sydney-Singapore-Paris is now QF233, while Sydney-Perth-Paris remains QF33. Eastbound routing remains unchanged.

From 18 May to 26 July, Paris increases to 5x/week and operating via Singapore in both directions, thus operating as Sydney-Singapore-Paris, and vice versa, excluding Perth.

To accommodate the additional B787 flying time on the Rome and Paris routes, Brisbane-Los Angeles will switch back to the A330-200 during the affected period, while some domestic A330 flying has been replaced with narrowbody aircraft to free up the A330s for Brisbane-Los Angeles. Flights to London will remain unchanged with the Perth-London flight continuing to operate westbound via Singapore, and non-stop on the return.

Apologies for any typos or inconsistencies, but we’re pushed this out a little quicker than expected! We’ll update it later!

The northerly routing adds 30 to 45 minutes to the flight time. This seems like a small amount of additional time, so why such a significant payload restriction necessitating a refuelling stop? It’s easy to forget that even at the best of times, QF9 is one of the world’s longest non-stop flights, pushing the capability of the B787-9.

It’s also not the first time Qantas have rerouted QF9, having done it twice in 2024 for similar reasons (12-27 April and 8-22 August). As with the previous times, the return leg (QF10) remains non-stop.

The logistics of adding a fuel stop mean that it can’t be implemented on an ad-hoc basis or scheduled at short notice. If it were to take an unscheduled stop it would certainly lead to the crew exceeding their flight duty limits, leading to the remainder of the flight being cancelled. This would lead to greater disruption for passengers and knock-on effects to the rest of the network with an aircraft and crew being out of position for at least 24 hours.

Scheduling the fuel stop in advance takes time as a replacement crew needs to be positioned to Singapore ahead of time. Singapore isn’t a regular Qantas B787 station, limiting its ability to utilise crew that may already be in position. Thus, despite the first airspace closures occurring on Saturday 28 February, Qantas only began the rerouted QF209 on Wednesday 4 March.

In the interleading time QF9 continued to operate non-stop, utilising the longer northerly routing, taking a payload penalty to carry additional fuel and/or lighted the load to reduce fuel burn. But just how big of a penalty is it?

Qantas were on record saying that introducing the Singapore stop would enable an additional 60 passengers to be carried. That’s a surprisingly large number for an aircraft that only seats 236 passengers, right? Were they really limited to just 176 passenger?

It’s actually missing some context! Firstly, we shouldn’t assume that QF9 was able to operate with a full passenger load prior to Saturday’s disruptions. Furthermore, we also shouldn’t assume that it can do this year round.

As usual, we brought the receipts! We’re able to estimate the payload restrictions that Qantas typically face on the route by analysing the declared seat capacity on QF9 and 10 from BITRE data.

Over the last year, the outbound seat capacity on QF9 (westbound Perth to London) averaged 219 seats per flight, while the inbound seat capacity on QF10 (eastbound London to Perth) averaged 235 seats per flight. We can interpret this to mean that they had to leave 16 seats empty on the average westbound sector, but just 1 empty on the average eastbound sector.

More importantly, we see significant variation in the payload restrictions throughout the year. Seat capacity was as low as 203 in January 2025, and as high as 234 in August and September 2025. There is also a clear seasonal pattern, with payload restrictions significantly more severe through the Australian spring and summer.

Back to our analysis, capacity was limited to an average 206 and 219 seats in February and March 2025. If we were to assume that this holds for February and March 2026, this would indicate a loss of 30 and 17 seats even before the rerouting. So the question becomes, how does the rerouting lose and additional 30 to 43 seats?

Thanks ACARS!

We have a single snapshot to show this in practice using the ACARS loadsheet for QF9 on 3 March. We don’t generally see Qantas’s loadsheets on ACARS since they’re typically not transmitted to the aircraft via ACARS (at least not utilising methods that allow us to see them). However, they tend to be transmitted by ACARS when flights are departing from outstations or when there are late changes. Basically, we got lucky!

We picked it at the time and referenced to it in some exchanges on social media, but we thought that it deserved more detailed analysis. QF9 on 3 March was the last non-stop to London before the rerouting via Singapore and it shows us just how challenging things were.

The loadsheet indicated TOB of 172 meaning “total on board” of 172. As an aside, during the Bad Bunny charter shenanigans a few weeks back some people misinterpreted this number as passengers on board. This isn’t passengers on board, but total, thus including crew. The flight typically has 16 crew member including 4 flight deck and 12 cabin crew, meaning that there were 156 passengers on board. Provided this was the full load that they could’ve carried it means the rerouting could carry 80 more, not 60, but we’ll give them some latitude here.

But this is where it gets really interesting! From the loadsheet we can also see that the estimated take-off weight (TOW) was 247t, well short of the aircraft’s maximum take-off weight (MTOW) of 254t. So did they leave 7t of payload behind? Well, no.

The loadsheet also tells us that take-off fuel (TOF) was 100.3t. The Boeing ACAP manual tells us that the B787-9’s physical fuel capacity is 126,429 liters or 101.5t. But does that mean they could’ve taken 1.2t more fuel? Again, no.

Fuel density varies with temperature. As temperature increases, the same weight of fuel takes up more volume. Perth is pretty hot at this time of year, and even at 7pm, temperatures are still in the low 30s (celcius). At 15C, 126,429 liters weighs 101.6t, but at 30C, the same volume weighs 100.1t.

Safe to say that the tanks were full. We’re just noting this so that people don’t critique the next part arguing that they could’ve taken more fuel. They couldn’t take more fuel, not because they were limited by weight, but because they were limited by volume!

So they were full of fuel with only 156 passengers on board and 7t short of MTOW. So why didn’t they take more passengers? Couldn’t they have increased the passenger load to increase payload by 7t and achieve MTOW?

We said Perth is hot at that time of the year, but even at 30C in still air, the B787-9 should lift nearly 250t, very close to its MTOW, and with just a small headwind it would lift MTOW. So it wasn’t take-off performance limits either.

The limits of absolute range

A few weeks back when we were analysing United’s new low density B787-9 configuration we received some criticism that our analysis didn’t take into account the impact of the B787-9’s forthcoming MTOW increase on its range.

We countered that the MTOW increase doesn’t increase absolute range because it doesn’t coincide with any increase in fuel capacity. While the MTOW increase is very useful for many missions it doesn’t change anything for missions that are already requiring the maximum fuel volume of the aircraft.

The increase in MTOW will increase the range at maximum payload, but this is already at a point where fuel has been traded for payload. The nuance here seems abstract, but its important. And this is where the rubber hits the road:

Fuel burn is a function of the weight of the aircraft. All else being held constant, a heavier B787-9 will burn more fuel than a lighter B787-9. Thus, had QF9 on 3 March carried more passengers it would’ve been heavier, burning more fuel. As we have already noted, it can’t carry more fuel, not due to weight limits, but the volumetric limit. Essentially, adding the extra flight time pushed the aircraft to its absolute range.

And that’s the real issue for QF9. It’s already at or so close to its absolute range. And further endurance needed, for example by taking a longer northerly routing can’t be made-up but trading payload to carry more fuel, rather it needs to reduce payload to reduce fuel burn. In some respects, it’s so close to the edge of its absolute range that its far more sensitive to the effect of a longer routing.

Some implications

Adding a fuel stop creates logistical challenges, but also may adds significant flight time, increasing the block time between Perth and London from from 17.5 hours to more than 20 hours. If sustained this may reduce the popularity of the flight with passengers, although it appears moot as long as significant capacity constraints exist at a market level with continued Middle East disruptions.

The positive externality is that it generates a significantly larger net gain in capacity. On the presumption that it can now carry a full passenger load it adds as much as 80 seats compared to the constrained longer non-stop routing. Comparatively, QF9 on 3 March lost between 50 and 63 seats compared to February and March averages. This is where Qantas is probably getting its 60 passenger number from, but as we can see, the number might be a little higher.

Another consideration is whether the Singapore stop would need to remain year round. As noted earlier, payload restrictions are less significant in the Australian autumn and winter. By August, near full passenger loads can even be carried on the typical non-stop routings. However, as we also noted earlier, this isn’t the first time Qantas have rerouted QF9, and one previous rerouting was implemented 8-22 August 2024, suggesting that Qantas will maintain the rerouting year round if the situation were to persist.

We hope this has been instructive, highlighting just how significant the payload restrictions become when adding an extra 30 to 45 minutes flying time to QF9, and why the Singapore stop is both necessary and useful at the current time given the significant supply constraint in the market.

Also, it goes a long way to show that the forthcoming increase in the B787-9’s MTOW just isn’t going to be much use for ULH flights since this isn’t the binding constraint on range.

It’s disproportionally impacted Australia given the importance of these airlines to international capacity. In the last year, they carried 4.6 million passengers to/from Australia, accounting for 10% of all passengers (12 months to September 2025).

The effects are particularly felt traveling between Australia and Europe, given the relative importance of their connecting capacity. Data presented by Virgin and Qatar during their ACCC JV application showed that they held substantial market shares between Australia and Europe. In the year ended May 2024, the ME3 accounted for 44% of the Australia-UK market, and 64% of the market between Australia and the rest of Europe.

The loss of this capacity is near catastrophic. Now into its third day, it’s affecting about 13,000 passengers a day just to/from Australia. The fleet is grounded, with many aircraft and crew “stranded” at outstations. Just in Australia there are 30 widebody aircraft on ground, with 9 at Melbourne and 8 at Sydney alone!

We’ve been getting lots of questions asking why don’t they just fly these aircraft direct to Europe, bypassing their hubs in the Middle East, utilising regional hubs in Southeast Asia instead? For example, people have proposed flying Sydney-Singapore-London, arguing that a large proportion of passengers on ME3 flights stuck in Sydney are headed to London.

While they don’t have the appropriate traffic rights under existing bilateral air service agreements, force majeure makes a case to allow such flights, even if just as ad hoc flight to allow stranded travellers to return home. While it would require some bureaucratic gymnastics to make it happen, it’s difficult to see governments preventing this. However, none of the airlines have indicated any interest or willingness to do this.

While it would be logistically difficult and very costly, it doesn’t really solve the problem. The strength of global connecting hubs is the fragmentation of traffic and diversity of destinations they serve, but it’s now their weakness. While a large proportion of those Sydney passengers might be headed to or from London, not all are, and those that aren’t are fragmented all over their networks, not just Europe. While it might get many passengers home, it won’t get anywhere close to all.

It would provide some relief for some passengers and send a signal of their willingness to mitigate the challenges that many are facing. However, the reason that they’re not doing it is more strategic and possibly for the greater good. It’s because this is their mitigation plan! This sounds ludicrous, right? Okay, but hear us out …

They’ve had months of warning and plenty time to plan it out. They’ve no doubt learned from previous airspace shutdowns over the last few years. Surely they could have come up with a plan to mitigate? Yes, and this is the plan!

Whether it’s the right contingency plan or it’s being well implemented, the plan is to keep the operation primed for a rapid restart. This is their focus, but to have a rapid restart they need aircraft and crew ready and waiting in the right places. And that’s exactly what they’ve done!

We can see this in action from the start. When UAE and Qatari airspace closed on Saturday the plan kicked into gear. There were a few key elementsn:

1. Flights departing from Abu Dhabi, Doha and Dubai: those that had departed

   would continue to their final destination, while flights that were awaiting departure from would be cancelled.

2. Flights returning to Abu Dhabi, Doha and Dubai from outstations: aircraft that had already departed from outstations to Abu Dhabi, Doha and Dubai would initiate diversions back to their point of origin.

3. Also flights returning to Abu Dhabi, Doha and Dubai from outstations: Aircraft yet to depart would not depart and remain on ground.

This strategy meant that very few aircraft and crew ended up stranded in the wrong places. Instead of having aircraft continue towards Abu Dhabi, Doha and Dubai in the hope that the airspace reopened by the time they got there, and ending up scattered all over the region, they now have all the aircraft and crews waiting at their point of origin, ready for a rapid restart ops at very short notice.

The success of this strategy is predicated on restarting operations in a matter of days. They can’t drag this out for weeks and months, and in some respects it’s a gamble. This explains why they haven’t implemented another plan, including a complex attempt to reroute flights bypassing their hubs since that’ll just end up with aircraft and crew scattered, slowing down or even breaking any quick restart.

It’s also why we’re not seeing forward looking cancellations. It’s noticeable that at no point have either of Emirates, Etihad or Qatar cancelled flights for an extended period, rather cancelling on a day-by-day basis.

For example, on Monday 2 March (at around 10am UAE time, 5pm AEDT), Emirates announced they were cancelling all flights to/from Dubai up until 3pm UAE time on Tuesday 3 March. This followed a similar announcement on Sunday 1 March, indicating that they were cancelling all flights to/from Dubai up until 3pm UAE time on Monday 2 March. Etihad and Qatar followed the same pattern.

Keeping it day-by-day allows them to restart at a day’s notice, hoping the situation improves sufficiently to allow a restart of flights. Keeping the aircraft and crew at outstations means they’re able to restart at short notice, while starting a bypass operation scatters crews and aircraft, and even if brilliantly planned and executed will end up with crew duty limits potentially impeding a restart.

But this strategy can’t go on indefinitely. You can’t go day-by-day for weeks or months. How long can it go on for? We simply don’t know!

Longer term trends are even more impressive. Over the last decade, travel between Australia and India has increased 81%, and has increased in relative importance having grown from 3.2% of all travel to/from Australia a decade ago to 4.9% in 2025.

The market is also relatively balanced with the Australian point-of-sale accounting for 57% of the market. Furthermore, the Australian point-of-sale has grown more rapidly in recent years, increasing 9% over the last year and 41% since 2019.

We’ve previously analysed the reasons for this growth, ascribing it to growing migration, cultural and economic links between the countries, spurred on by India’s strong economic growth.

The growth has contributed to a large build-up in direct flights. In 2019, Air India were the only airline providing direct flights, operating 3x/week Delhi-Melbourne and 5x/week Delhi-Sydney flights (both B787-8). Since then, both routes have increased to daily, while Air India also introduced 3x/week Mumbai-Melbourne (since suspended due to aircraft availability challenges). From July 2026, Air India will upgrade Delhi-Melbourne to the B777-300ER, resulting in a significant increase in capacity.

In 2021, Qantas reentered the Indian market after nearly a decade long absence with seasonal Melbourne-Delhi (3x/week) and year-round Sydney-Bengaluru currently operating up to 6x/week.

Despite the large and growing market, passengers on direct flights make up a small proportion of the market. In 2024, BITRE indicates 301,934 arrivals on direct flights from India, while ABS indicates 992,730 total arrivals from India, meaning that net transit traffic accounts for 70% of total travel between the two countries (BITRE data is not yet available for 2025).

We’ve previously argued that this is due to the fragmentation of the market, particularly on the Indian side, as well as the seasonality of the market. But what does this mean? We’ve come across some detailed data on O&D city pairs between Australia and India that helps us explore this in more detail …

Seasonality

Australia-India is big market, with an average of 2,879 inbound passengers (to Australia) per day in 2025. To put that in perspective, that’s nearly 6 of Qantas’s A380s or more than 8 of Air India’s B777-300ERs each day. But daily averages estimated from annual aggregates can be deceiving, and even misleading!

The market is incredibly seasonal with significant seasonal peak in the southern hemisphere summer, coinciding with the Christmas holiday period, likely an artefact of the dominance of outbound VFR traffic. Furthermore, the seasonality varies between inbound and outbound traffic.

Data from 2025, shows that the busiest month for arrivals in Australia was January with 154,310 passengers, compared to just 58,560 in the quietest month (July). That amounts to 4,978 inbound passengers per day in January (nearly double the annual average), but just 1,889 passengers per day in July.

Seasonality generates significant challenges for airlines offering direct flights, and likely contributes to the large proportion of traffic utilising 3rd country transit hubs like Malaysia and Singapore. These hubs are able to balance seasonality with other routes, for example, connecting Australians to Europe during the northern hemisphere summer, or more stable business traffic between Australia and Southeast Asia throughout the year.

We’ve previously analysed how Qantas and Air India manage this seasonality, with Qantas varying frequency and operating seasonal routes to reduce capacity during the low season.

Air India’s capacity is more consistent, with an emerging strategy focusing on connecting traffic between Australia and Europe during the low season. This is an elegant strategy given the counter cyclicality between Australia-India and Australia-Europe, with the latter peaking during the Australia-India low season.

Meanwhile, Qantas pivot that capacity directly towards Europe rather than through India, for example, their seasonal Perth-Rome flight. Take a look out our detailed analysis of this from last year to see this seasonality in more detail. Seasonality is a contributing factor to the dominance of 3rd country connecting traffic between Australia and India, but it’s not the primary cause. We’ve hypothesised that its the fragmentation of the market. But what does this mean?

Fragmentation

The BITRE data that we utilise as the core of many of our analyses is great at measuring capacity and passenger flows on direct flights, but it doesn’t say anything about network traffic. For example, we know that Malaysian Airlines carried 1,015,891 passengers to/from Australia in 2024, and while we know anecdotally that they’re a major player in the Australia-India market, we can’t estimate how many of those 1,015,891 passengers were carried between Australia and India.

While we can triangulate the BITRE data with ABS data, it still only generates aggregate estimates, for example, the previously noted estimate that 70% of Australia-India traffic transits via 3rd countries is a “net” measure. We don’t know where they’re transiting through, or connecting behind or beyond. What we really need are estimates of O&D city pairs, however these data aren’t available from public sources and only available from fairly expensive proprietary databases.

Fortuitously, a short analysis by Behramjee Ghadially on LinkedIn a few months back shared O&D city pair data for the largest 16 city pairs between Australia and India (8 largest Indian and 2 largest Australian cities) in 2024. These data count both direct and transit traffic, measuring both directions of travel.

The O&D city pairs indicated a total demand of 1.42 million passengers in 2024. Triangulating this with the ABS data analysed earlier, the 16 busiest city pairs accounted for approximately 71% of the total Australia and India market. Note that the ABS data only measures arrivals, but scaling either series allows triangulation.

Of the 16 city pairs, only 3 currently have direct flights, although a 4th did operate during 2024 (the now suspended Mumbai-Melbourne). Nevertheless, direct flight city pairs accounted for only 34% of the total market (38% including Mumbai-Melbourne).

Turning this around, it means that routes without direct flights account for 66% of all traffic between Australia and India (62% if Mumbai-Melbourne were included). If a new direct flight were added to the largest unserved city pair (i.e. Hyderabad-Melbourne) this would only decline to 57%.

We can break this down on both sides. On the India side, Delhi accounts for 29% of the total market, while Bengaluru, Mumbai and Hyderabad account for just 8% each. Combined, the 2nd, 3rd and 4th largest markets are smaller than the Delhi market.

As a methodological point, this is actually an undercount since Delhi’s 29% only refers to the 71% accounted for by the 16 busiest city pairs. For example, Delhi-Brisbane would be included in the “other” category. This is evident when considering the same figure on the Australian side and observing the “other” category is also 29%. Nevertheless, on the Australia side, the market is dominated by Melbourne and Sydney with 40% and 31%, respectively.

This is what we mean by fragmentation: that the traffic between Australia and India is spread across so many cities, particularly on the Indian side. What these figures are showing us is that the fragmentation is far more extensive on the Indian side, with at least 71% of the Australian side dominated 2 cities. This goes a long way to explain how both Air India and Qantas have concentrated their direct flights on Melbourne and Sydney.

While the Indian side is fragmented, requiring 8 cities to cover the same dominance as 2 on the Australian side, Delhi is an outlier as it accounts for a disproportionately large share of the traffic on the Indian side. This also goes a long way to explain why both Air India Qantas have concentrated a lot of direct capacity on Delhi. The outlier is Qantas’s Sydney-Bengaluru flight!

While Sydney-Bengaluru is the 3rd largest city pair, it accounted for just 3% of the total market of 95,726 annual passengers or just 131 passenger per day in each direction. No doubt, the Hyderabad fanboys and fangirls are reading this and getting ready to rumble! How is it that a city with at least 156,664 annual passengers between Australia and India doesn’t have direct air services? That’s 215 passengers per day each way, more than enough for a 3x/week flight, right?

The challenge lies with seasonality, yields and connecting traffic. If you were wondering why we focused so much of the earlier part of the analysis on seasonality it’ll show now.

Seasonality revisited

We can’t breakdown the O&D city pairs by month, but we can apply aggregate parameters based on the seasonality in the ABS data at the route level. While this assumes there is no variation in seasonality by city pair, it’s instructive nonetheless.

Let’s take the largest unserved city pair, Hyderabad-Melbourne with 85,581 annual passengers and an average of 117 passengers per day each way in 2024. Obviously, not every route requires a daily flight, so consolidating demand weekly means means demand averages 821 passengers per week in each direction. A 3x/week flight on Air India’s B787-8s would supply 768 seats, while the same on Qantas’s A330-200s would supply 753 seats. This seems viable, right?

Not so fast! When applying the aggregate seasonal proportions at the route level, we see the effect of the seasonality. While January generated demand for 196 passengers per day in each direction, July only generated demand for 83 passengers per day. On a weekly basis that’s 1,374 passenger per week in each direction in January, but just 578 passengers per week in July. So even a 3x/week flight would run into problems in the low season.

One might question why we don’t add Hyderabad-Sydney to this, but that belies the point since there’s no network advantage for a passenger flying Hyderabad-Melbourne-Sydney compared to Hyderabad-Bengaluru-Sydney, Hyderabad-Delhi-Sydney, or any number of 3rd country transit options (e.g. via Kuala Lumpur, Singapore, Bangkok, Hong Kong, etc). And this is the fundamental challenge for airlines in fragmented markets.

This means that home market carriers like Air India and Qantas can only generate a network advantage for the non-stop O&D city pairs. Seasonality compounds this by reducing the market size of many O&D city pairs, and significantly undermines the viability of many Australia-India city pairs.

Even looking at the largest O&D city pair (Delhi-Melbourne) we can see the challenge it presents. With annual demand for 363,192 passengers, this translates into an average of 498 passengers per day in each direction. January averages 833 passengers per day in each direction, while the weakest month (July) generates demand for 351 passengers per day. On a weekly basis, that’s 5,833 passengers per week in January, and 2,455 passengers per week in July.

We’ve noted how Air India are increasingly managing seasonality by pivoting capacity to onward connections to Europe (more on that later). However, Qantas only operate Melbourne-Delhi seasonally, from the end of October through the end of March. If Qantas can’t maintain Delhi-Melbourne annually, it’s unlikely that Air India could maintain Hyderabad-Melbourne year round without the balancing connecting traffic onto Europe. And since Air India don’t serve any European destinations from Hyderabad this isn’t viable. So why not seasonal like Qantas’s Delhi-Melbourne flight?

Generally speaking, seasonal flights generate higher operational costs and will have shorter periods over which to share the fixed costs. To overcome this, airlines must earn very good money during the high season. But not all routes have the same earning potential and just because a route has solid demand, it doesn’t follow that the demand is at the price points required to make the route a success.

Follow the money

Passenger demand is one thing, but what are customers willing and able to pay? It’s all good and well arguing that Hyderabad is a bigger market than Bengaluru, but if airlines can generate higher prices from Bengaluru then they’re probably going to focus capacity on Bengaluru over Hyderabad. So what does the data say?

Data shared by some of our friends on social media highlight just why Qantas probably chose to fly Sydney-Bengaluru rather than to Delhi, Mumbai or Hyderabad. It shows that Sydney-Bengaluru had the highest average fare of Sydney-India city pairs during FY2024, earning an 8% unit revenue premium over Mumbai and Chennai, and a larger 25% premium over Hyderabad. One might counter that Sydney-Delhi earns the lowest prices, however that doesn’t account for the relative size of the market. Delhi-Sydney is more than double the size of the 2nd largest (Sydney-Bengaluru), and larger than Mumbai, Hyderabad and Chennai combined.

Qantas must chose between higher price or larger market, while also considering competition. For example, while Sydney-Delhi is larger, it has competition from Air India’s non-stop flight. Presented with the choice between Bengaluru or Hyderabad, Qantas were likely swayed by a larger revenue potential for a similar market size. Another consideration are the network effects and connecting traffic.

Other data shared on X shows that only 41% of the traffic on Qantas’s Sydney-Bengaluru flight in 2024 was point-to-point traffic (i.e. between Sydney and Bengaluru), while 59% connected at one or both ends. Of the 10 biggest connecting markets, 6 were on the Indian side (Hyderabad, Mumbai, Pune, Ahmedabad, Cochin and Delhi) and 3 on the Australasian side (Melbourne, Brisbane, Auckland), and 1 further beyond Sydney (Honolulu).

Of the 59% of traffic that connected, data from a comparable source shows that approximately 34% connected at Bengaluru and 25% at Sydney. This once again highlights the fragmentation on the Indian side. Sydney is Qantas’s largest hub - this is not in question - however, is Bengaluru an optimal connecting hub for Qantas?

Geographically, points further south and east in India reduced backtracking when flying to/from Australia, but that doesn’t mean they’re the most efficient options. Qantas must consider the number of destinations served from each city, available capacity and alignment of schedule. Furthermore, they must consider these in the context of their codeshare partner, IndiGo.

Looking at Air India and IndiGo’s domestic networks provides some insight into Air India and Qantas’s choices. Once again, the data are imperfect. We don’t have access to proprietary data that would allow us to measure seat counts for each carrier at each airport. A recent article by Network Thoughts gives us a snapshot for a single month (August 2025), showing that approximately 33% of Air India’s seat capacity (including Air India Express) was concentrated on Delhi, reinforcing Air India’s choice to concentrate Australia-India capacity on routes to/from Delhi, despite its geographic location.

Meanwhile, IndiGo doesn’t concentrate their capacity anywhere nearly as much as Air India, with only 20% at Delhi. But proportions can be deceiving since IndiGo’s have 27% more seats to/from Delhi in absolute terms than Air India, despite its lower relative share. Given IndiGo’s more varied capacity distribution and their relatively larger scale, Qantas’s have more choice than Air India. While Bengaluru and Hyderabad would probably both work well as a hub for Qantas, they have the luxury to chose the point which will give them higher fares on O&D traffic, while generating similar connectivity from an efficient geographical position.

We’re able to validate this with an alternative metric. Public data from Indian authorities allows estimation of the number of flights by each airline to/from each airport. It’s imperfect as it doesn’t account for variation in aircraft size, but it’s a good proxy for capacity nonetheless. Another of our friends on X does a great job of presenting and visualising these data, telling a similar story.

The varying domestic networks highlight how and why Air India and Qantas have developed different strategies, with Air India focusing their capacity on Delhi, despite its relatively inefficient location. While Delhi provided Air India with greater critical mass for domestic connections, it also provides them with the largest O&D markets and critical onward connections to Europe.

Meanwhile, Qantas have picked a destination with sufficient domestic connections and an optimal geographic location alongside the most lucrative ticket prices. And unlike Air India, Qantas have no interest in onward connections beyond India (i.e. to Europe)! This is a great example of how two airlines can adopt different strategies on overlapping routes. In both cases, they’re sound strategies.

Implications

But what are the implications of this all? How will the fragmentation and seasonality of the market affect future expansions of Australia-India flying?

Both Air India and Qantas have signalled their intent to expand capacity between Australia and India. Air India added Mumbai-Melbourne in 2024, although this didn’t last. They’ve since increased planned capacity on Delhi-Melbourne by switching the B787-8 for the larger B777-300ER. Meanwhile, Qantas International CEO Cam Wallace has spoken publicly about expanding capacity to India, highlighting that they’d prefer to do this with B787s rather than the current A330s.

If we’d considered this question a year ago, we’d have answered this very differently to how we would today. We wouldn’t have been wrong, however Air India have evolved their India-Australia strategy. In 2024, they were adding Mumbai-Melbourne, but in 2026 they’re increasing capacity on Delhi-Melbourne. More importantly, they’ve retimed both their Delhi-Melbourne and Delhi-Sydney flights, with an emphasis on onward connecting traffic to Europe. We’ve hypothesised that they’re looking to balance seasonal variations in the Australia-India market with the countercyclical Australia-Europe market.

Network Thoughts explained this very clearly at the time, highlighting how the retiming enabled passengers from Melbourne or Sydney to have a 65 to 80 minute connection to Frankfurt and Paris, although there are question marks regarding the return flights that don’t make viable connections (same for Amsterdam, Birmingham, Manchester, Milan, Rome, etc). London is the only European route that has efficient connections to/from Melbourne and Sydney on Air India’s network in both directions.

So we need to narrow the hypothesis: it’s not Europe, it’s London! Would they really do this for just one destination? Well, yes! The Australia-UK market is huge! To put it in perspective, it was nearly 30% larger than the Australia-India market was in 2025, accounting for 1.4 million inbound passengers per ABS data. Australia-UK accounted for 39% of all Australia-Europe travel, and 46% of the outbound Australian market.

Given this strategy, what does Air India do? Do Air India add new points in India (e.g. Mumbai, Bengaluru or Hyderabad) to connect to/from the larger O&D markets of Melbourne and Sydney, or do they seek to add more points in Australia to connect to Delhi (e.g. Adelaide, Brisbane or Perth)?

If they’re looking at building the connecting model then surely they’ll look to add another point in Australia and connect it to Delhi! And just this week, The Hindu reported that Air India are considering just this, with Adelaide and Brisbane being considered.

However, they also note that “the proposed Australian services may be operated from Mumbai and Bengaluru, in line with the airline’s strategy to strengthen multiple metro hubs and distribute long-haul operations beyond Delhi.” While this may under consideration, we’d be very surprised. Ultimately, if Mumbai or Bengaluru were being considered on the Indian side, we’d fully expect this to connect to Melbourne or Sydney given the larger O&D markets. If Adelaide or Brisbane are being considered, then we’d fully expect those to connect with Delhi to enable those onward connections and balance the seasonality.

On the other hand, it’s a little more intriguing on the Qantas side. If they were to open up new Indian stations, we’d fully expect that to focus on Sydney and possibly Melbourne, i.e. Melbourne/Sydney-Hyderabad/Mumbai. However, these options would be hampered by seasonality and it’s questionable whether they Hyderabad or Mumbai would generate sufficient yields to justify a seasonal service. Alternatively, they could look to add additional capacity from other points in Australia to Bengaluru, increasing the scale and critical mass of their existing operations.

This sounds very ambitious, but before Air India or Qantas open new points on the Australian size, they must consider the relative demand. A significant driver of the market is the growing Indian diaspora in Australia. The 2021 Australian census shows that this community is heavily concentrated on Victoria and New South Wales, hence Melbourne and Sydney. Of the Indian born population in Australia, 272k and 219k resided in Victoria and NSW, respectively, whereas Queensland (Brisbane) and South Australia (Adelaide) were home to just 77k and 47k, respectively.

The relative scale reinforces how much smaller the Brisbane and Adelaide markets are, and how much more dependent they’d be on onward connections to Europe or London. As such, we’re skeptical of Air India wanting to start new routes from Adelaide or Brisbane to Mumbai or Hyderabad, rather than to Delhi.

Thanks for reading, and remember to subscribe, and share with friends and colleagues.

Share

Melbourne-Sydney alone accounted for 15.0% and by any metric is one of the busiest and most lucrative routes in the world. While substantially smaller, Brisbane-Sydney and Brisbane-Melbourne are still the 2nd and 3rd largest routes in the country, accounting for 7.9% and 5.9%, respectively.

RPT refers to scheduled airline services in Australia that are available to the public for purchase. This excludes charter flights, including FIFO mining charters that are a significant portion of air travel in Australia. FIFO flights are not available for sale to the public and rarely serve major routes, rather serving remote mine sites.

The dominance of triangle isn’t surprising given the concentration of population around the physical triangle forms by the three cities. The broader metropolitan areas of these cities accounted for 51% of Australia’s population. In fact, their dominance in Australia’s population results in the “center of population” lying just to the west of the “triangle” near the town of Ivanhoe, NSW.

What is surprising though is how the dominance of the triangle has waned in the aftermath of the COVID-19 pandemic. This wasn’t a slow or consistent decline, but what appears to be a structural downward shift.

In the 12 months to October 2025, the triangle represented 26.7% of domestic passengers, with a relative peak of 26.9% in April and May of 2025. This is significantly lower than the the 28.7% in 2019, and the relatively peak of 29.0% observed for several months in late 2017 and early 2018. Not only is the current proportion lower than the pre-pandemic period, it’s the lowest recorded since the late 1980s, before the deregulation of the Australian domestic airline market.

Read more

It’s somewhat unusual for a European network carrier not to develop their primary hub in their largest O&D market, like British Airways at London Heathrow, Air France at Paris CDG or Iberia at Madrid. One of the primary reasons that Lufthansa’s didn’t develop Berlin as their primary hub in the post-WWII era was its dislocation from the rest of West Germany, and Berlin’s split into West and East Berlin.

Geopolitics isn’t the only reason, and there’s a strong argument that even in a unified post-WWII Germany that Frankfurt would’ve served as Lufthansa’s primary hub anyway. Frankfurt served as a major logistics hub prior to WWII and the fragmentation of Germany’s population across a larger geographic area means locating logistics hubs closer to the geographic centre of the country made more sense. To put this in perspective, London accounts for 13% of the UK population, while Berlin only accounts for 5% of the German population.

Yet, Frankfurt was the 22nd busiest airport in the world, and the 5th busiest in western Europe, handling 59.3 million passengers in 2024. Given that it’s only Germany’s 5th city, who the hell is actually flying there? Well, the answer is quite simple, most of those 59.3 million passengers are actually just connecting through Frankfurt.

Analysing connecting hubs is really difficult since few airlines share sufficient similarities in strategies and economic models to make a like-for-like comparison. Lufthansa gives us a unique opportunity due to their split hub model. While Munich is a much smaller hub, it’s still substantial, handling 31.6 million passengers in 2024 and making it western Europe’s 8th busiest airport.

This allows us to make a like-for-like comparison between Lufthansa’s operations at Frankfurt and Munich. One interesting point of comparison is the proportion of O&D versus connecting traffic they’re carrying on overlapping routes (i.e. routes which are operated from both Frankfurt and Munich, e.g. Frankfurt-London and Munich-London).

Read more

The B787-9 is already a major part of United’s widebody fleet with 48 of the type, accounting for 20% of their widebody fleet. In the coming years it’ll become the mainstay of their fleet with at least 141 more on order, slated to replace all of United’s B767, a large chunk of the B777s and providing the opportunity for significant incremental growth.

United’s contemporary B787-9 configuration is already relatively premium heavy, seating 257 passengers in a 3-class layout with 48 business, 21 premium economy and 188 economy class seats. United further segment economy class into economy plus and economy, however this is a branding exercise for seats with extra legroom - there’s no need to consider it as a seperate cabin.

Meanwhile, the new configuration is even more premium heavy, seating 222 passengers in a 3-class layout with 64 business, 35 premium economy and 123 economy class seats. The configuration trades 65 economy class seats for 16 more business class and 14 more premium economy seats. Not all B797-9s will be configured in this very low density version, with United committing to at least 30 aircraft by 2027, likely from new orders rather than reconfiguring existing aircraft.

The commentary on social media is reacting as if this is extraordinary or unusual. While the new configuration does represent a significant reduction in cabin density compared to their contemporary configuration, it’s not extraordinary. In fact, several airlines equip their B787-9s with similar, and sometimes even lower density cabins. For example, Japan Airlines and ANA both have configurations with fewer seats, while Air New Zealand’s forthcoming low density configuration is relatively similar.

But this is somewhat besides the point, as the commentary has missed why United are implementing this new configuration. For example, Kyle Potter from Thrifty Traveler questioned whether it’s an overcorrections to the boom in premium travel, while Zach Griff argued that “United is betting big on premium demand sticking around”. We don’t think it’s an overcorrection or a big bet. Instead, we think it’s about payloads, payload restrictions and United optimising their cabins for their now large ultra long haul (ULH) network, something they hadn’t necessarily contemplated when their inducted the B787-9 in 2014.

United’s ULH network

United utilise the B787-9 on some incredibly long missions, many of which are relatively new routes for United. Of United’s ten longest routes, nine are operated by the B787-9, with only Los Angeles-Sydney operated by B777s. Five have been inaugurated since United’s 1st B787-9 was delivered in 2014. This includes the four longest routes. The other five were previously flown on other types (particularly the B747-400 and B777-200ER), and have since been replaced with the more efficient B787-9.

This highlights how the B787-9 has enabled United to open several new ultra long haul (ULH) routes, partly because of the B787-9’s longest range and/or the improved economics at these distances. Critically, United have pushed the range of the B787-9 over time, beyond its initial business case.

However, ULH routes come with significant operational trade-offs in the form of payload restrictions that limit the number of passengers they can carry on each flight. Payload restrictions vary between each route, with significant seasonal variation due to variations in the jet stream and other seasonal weather conditions. Notably, westbound routes are more significantly affected due to the prevailing jet stream.

We can glean a lot into payload restrictions on individual routes using USDOT data that allows us to estimate the variations in available payload between westbound and eastbound flights on the same route on a monthly basis. This is instructive since westbound routes are longer than eastbound routes due to the prevailing jet stream, and as the distance of that the flight covers increases, so the absolute variation in time to fly the westbound and eastbound sector increases.

Available payload on United’s B787-9 flights

Lets have a look at the available payloads on United’s longest B787-9 routes, using San Francisco-Singapore as our primary example. While it’s not United’s longest B787-9, it’s the longest year round and allows us to example how payload restrictions vary over the year.

Over the last year, the westbound sector of United’s San Francisco-Singapore flight averaged an available payload of 27.9t (metric tons), 4.2t less than the 32.1t available on the eastbound sector. This highlights the variation between west and eastbound sectors due to the prevailing jet stream. This is in addition to the seasonal variation, with much larger differences between west and eastbound sectors in the northern winter months between November and April.

The pattern is consistent for most ULH routes, although the magnitude varies route by route. An exception were both Cape Town routes, which might be ascribed to the routing covering a greater north-south rather than east-west distance. Some of the widest gaps were 8.7t for Houston-Sydney and 9.3t for Johannesburg-Newark. In the case of the former, the flight only runs during the northern winter, and thus biasing the sample. It’s for this reason that we focus more attention on San Francisco-Singapore rather than United’s longest B787-9 routes.

In the case of the latter, the variation is due to take-off performance from Johannesburg. While the route still has variation across the year (e.g. the difference is as small as 5.6t in August 2025, and as large as 13.3t in December 2024), the absolute gap is affected by the aircraft’s reduced take-off weight at Johannesburg due to its “hot and high” conditions. The airport is located at approximately 5700ft above sea level, significantly limiting the aircraft’s takeoff performance.

How does this translate into payload restrictions?

The B787-9 has a maximum payload of 52.5t that can be utilised for a combination of passengers and cargo. Rarely will an airline utilise all the available payload as its passenger and/or cargo load will rarely be big enough, or even max out on volume before weight. Meanwhile, on longer flights the available payload will be significantly lower than 52.5t as the aircraft trades some of this payload to carry more fuel.

This might not be obvious, but this trade-off is key to the flexibility of the aircraft, allowing airlines to operate a variety of missions. The B787-9 has a maximum takeoff weight (MTOW) of 254.7t and a maximum zero fuel weight (MZFW) of 181.4t. The MZFW includes its maximum payload of 52.5t, meaning that at maximum payload it’ll only be able to carry 73.3t of fuel, well short of its maximum capacity of 101.4t of fuel. Meanwhile, if it carries its maximum fuel of 101.4t, it means it’s payload will be limited to just 24.4t.

The point at which this trade-off starts to occur varies by aircraft, however ostensibly the B787-9 will start to trade payload for fuel for routes beyond 5300nm (approximately). This varies day-to-day, route-by-route, but as previously noted, United’s ten longest B787-9 routes are all longer than 6364nm, and several significantly longer, meaning that available payload will be significantly less than 52.5t.

Going back to the San Francisco-Singapore route, over the last year it averaged an available payload of 27.9t and 32.1t for west and eastbound departures, respectively. This means that the westbound and eastbound departures gave up an average of 24.6t and 20.4t of payload in order to carry the required fuel for ULH operations. Recalling that we noted that few routes need to carry the maximum payload, lets consider the practical implications.

Practical implications

Taking United’s contemporary 257 seat configuration, we can estimate the weight of a full passenger load. The standardised FAA estimate is 93-96 kg per person (205-210 lbs), depending on the season. This include carry-on baggage, but not checked luggage. Assuming a full passenger load and an average of 20kg checked luggage per passenger this would amount to a payload of 29.1-29.7t, higher than the 27.9t average available on the westbound San Francisco-Singapore departure.

Not only is this higher than the annual average, it’s higher than the average for 9 of the last 12 months, and as much as 4.7t more than the available payload in the most constrained month. On the average westbound sector this would require a reduction in passenger load of 15 or 16 passengers, however this would be as much as 40 passengers on each westbound sector in the most constrained month (this is itself is an average of the flights during that month with some flights potentially even more).

Houston-Sydney and Johannesburg-Newark will require payload restrictions on the average flight, while Los Angeles-Melbourne will require payload restrictions for most months of the year, and both Cape Town routes for nearly half the year. However, this assumes zero cargo, and any cargo carried will increase the need for payload restrictions.

Furthermore, this only considers westbound routes, with some eastbound ULH sectors also requiring passenger payload restrictions, although during fewer months of the year.

Let’s put ourselves in United’s shoes for a moment

Instead of 257 passengers, your average westbound San Francisco-Singapore sector would be limited to 241 or 242 passengers over the whole year and significantly lower in many month (as low as 217 in the worst month). This would be worse if they needed or wanted to carry cargo.

Somewhat obviously, the airline would want to limit the payload restriction to economy class to minimise the lost revenue from the restriction. And if you operate enough ULH flights subject to payload restrictions like these it might make sense to operate a subfleet with fewer economy class seats. If they can’t be sold then why have them on the aircraft in the first place?!? Not only does it affect revenue but also costs as a restricted passenger load effectively increases the unit cost (cost per available seat mile or kilometer) as costs are shared over fewer available seats.

As simplistic as this sounds, this is exactly what they’ve done. They’re reduced the number of economy class seats by 65 (188 versus 123) on the new low density configuration. While this number is larger than what we might expect because it’s a static metric. Since they’re removing economy class seats, they’ll have more floor space available for more premium seats (business and/or premium economy), meaning that we should be focused on the net change, which is 35 fewer seats (257 versus 222).

The 35 seat net reduction means that a full passenger load would amount to a payload of between 25.1 and 25.6t, slightly less than the 27.9t available on the average westbound San Francisco-Singapore sector. This would requiring a negligible passenger payload restriction during only one month of the year! It’s almost as if they calibrated the configuration to this route!!!

Now United will carry the same number passengers on the route, while the configuration will enable them to earn more revenue while doing so, with more premium seats available for sale. The timing of this discussion is very relevent given our recent analysis of Qantas’s Project Sunrise and the revenue and cost challenges, and opportunities, flying ULH routes.

Conclusion

Firstly, the very low density version is only a subfleet, with at least 30 aircraft by 2027. Expecting that these will be new builds, and not retrofits, this will account for, at most, 38% of the B787-9 fleet. It certainly won’t be the predominant configuration.

Secondly, and most instructively, they’ll debut on San Francisco-Singapore in early 2026. This is their 2nd longest B787-9 route, but their longest year round route as Houston-Sydney only operates in the northern winter season and will end its seasonal run before the new configuration is enters into service. We wouldn’t be surprised to see the new configuration on Houston-Sydney next year.

Its entry into service on the longest route on the network is strongly indicative of its purpose: to compensate for the higher unit costs imposed by payload restrictions on ULH flights with higher unit revenues enabled by larger premium cabins! If you’re going to leave seats empty on enough routes, it’ll pay you to just remove those seats and if you have more space, then it’ll also pay you to utilise that space for higher yielding passengers.

San Francisco-Singapore is a twice daily route and will require the at least 4 aircraft, and shifting the ten longest routes to the new configuration would require more than 20 aircraft. This is still less than the 30 aircraft fleet that United have indicated they plan to operate. The higher number likely reflects several possibilities including further growth in ULH flying, efficient scale of the subfleet to ensure optimal utilisation, and the potential to utilise it on several niche high-yielding routes that may not warrant the configuration for its range capabilities. All have a likely degree of validity.

The latter point is also important since United have already indicated they also plan to schedule the aircraft on various London Heathrow services. Scheduling some shorter schedules will also help ensure high utilisation of the subfleet.

We really enjoyed this analysis since it’s a great illustrative example of the operational constraints of an aircraft. Too often we hear things like the B787-9 has a 7500nm range (this is its advertised range). Indeed, the aircraft can fly that far - just as we’ve seen here - but it’s not a generalisable figure since this simply doesn’t mean that it can fly a full payload, nevermind a full passenger load, this far.

In many cases this isn’t a practical constraint for airlines, however as an airline stretches into ULH territory it becomes a rather dynamic constraint. United’s solution is actually very typical, comparable to other airlines like Air New Zealand and Qantas flying the B787-9 on ULH routes.

Thanks for reading and we hope you enjoyed the analysis. If you liked our analysis, then please share it widely!

Share

Many observers initially labeled Project Sunrise a publicity stunt. Even Cam Wallace, currently the CEO of Qantas International but then an executive at Air New Zealand, initially wrote it off as one. Qantas are very, very good at PR, so it’s easy to be cynical about it. However, recent history has shown Qantas’s ambitions to push the envelope on ultra long haul (ULH) flying. In the last decade, Qantas have launched several new ULH routes, each significantly longer than their previously longest flight from Los Angeles to Melbourne. Qantas’s push into ULH flying was enabled by their acquisition of the B787-9 that allowed them to extend flight times by nearly two hours.

Route     Distance Block*  Start
PER-LHR   7829nm   17:50   2018
DFW-MEL   7814nm   17:40   2022
PER-CDG   7702nm   17:30   2024
JFK-AKL   7671nm   17:40   2023
DFW-SYD   7454nm   17:25   2014**
PER-FCO   7211nm   16:25   2022
LAX-MEL   6883nm   15:55

* Block time of longest/westbound leg.
** Prior to 2014, westbound leg was operated via BNE.

In some respects, it’s not surprising to see Qantas pushing the envelop once again. We wanted to understand why they’re doing this? What’s motivating them in pushing block times to 21 or 22 hours that’ll be required for the westbound Sydney-London sector?

We’re not interested in how people feel about it, rather we’re interested in cutting through the nostalgia, resistance to change and spin, to understand the strategy and economics behind it. There’s a lot more nuance than meets the eye, so we wanted to challenge some of the simplistic narratives and understand the costs and risks involved, and the prize.

This analysis is very long, so apologies in advance, but there are a range of factors, and a whole lot of history that needs to be understood to help us reach an incredibly provocative conclusion: the tyranny of distance and the inability to compete on cost with one-stop carriers means they don’t have much of a choice other than a “deal with the devil”!

How people feel about it

Didn’t we just say we’re not interested in how people feel about it? We did, but let’s allow ourselves a short digression to reinforce why “how people feel” about it isn’t interesting or important. We know this sounds condescending, but bear with us.

Social media is a cesspool of engagement farming these days and Project Sunrise has been perfect fodder for asking people how they feel about 22 hour non-stop. The replies are just what you’d expect, with a lot of strong feelings about not wanting to do it. The algorithm will boost negativity and have us believe that the overwhelming opinion is that people hate the idea.

We’ve seen this before though. For those of us old enough to remember, there was a time when most long haul flights required several fuel stops. These have declined over time as aircraft technology improved. Old hands remind us how people bemoaned when flights like Melbourne-Los Angeles or Sydney-Johannesburg went non-stop following the introduction of the B747-400. Or how the 2nd stop was dropped from Sydney-London (and presumably the 3rd, 4th and 5th stops before that). These are now a distant memory, and most contemporary passengers would cry foul if any airline suddenly reintroduced a fuel stop in Fiji or Mauritius!

Los Angeles is no longer the frontier either, with non-stop flights to the United States now targeting more efficient connection hubs further afield (e.g. Dallas and Houston), providing passengers with more efficient connections to a wider range of destinations in the US. Really, who the hell wants to transit at LAX if you can avoid it?!?!

Nostalgia is powerful aphrodisiac and it’s hard to change the mind of someone who “feels” strongly about it. Qantas continue to shift capacity from Los Angeles and San Fransisco to Dallas, while American (Dallas) and United (Houston) have followed suit. We can only suppose they’re doing this because they’re following what customers want rather than strong feelings on social media.

As an aside, have a look at a short “chart of the week” we published earlier this year showing how patterns between Australia and the US have evolved over time.

More importantly, we’ve also seen a comparable evolution of flights between Australia and Europe, beginning with Qantas shifting the stopover point on their Melbourne-London flight from Singapore to Perth in 2018. Technology played a big role in this with the introduction of the B787-9 allowing Perth-London to be flown non-stop. The Perth-London sector has been so successful that Qantas dropped the Melbourne leg in 2024.

Our analysis at the time showed that the share of Melbourne passengers declined consistently as the non-stop Perth-London leg became more popular over time. By 2024, Melbourne O&D passengers accounting for one-fifth of the total passengers while the flight boasted load factors of well above 90%. Quite frankly, nobody predicted this was an outcome in 2018.

Qantas have since introduced other non-stop flights from Perth to Europe: a seasonal 3x/week Rome flight and more recently 3x/week to Paris. The success of Rome has led Qantas to announce a 4th weekly frequency and a longer operational season in 2026.

Feelings and nostalgia aside, it seems the median Qantas passenger either prefers these non-stop flights or at the very least is ambivalent about it given the plethora of alternate options, whether that be one-stop flights to Europe or connecting through Los Angeles or San Francisco.

Putting feelings aside, what is Qantas’s strategy?

Some question whether or not Project Sunrise is a huge risk for Qantas. They’re spending a lot of money on at least 12 purpose built and highly customised aircraft. The more customised an aircraft, the greater their residual risk (i.e. risk that the resale value of the aircraft declines very quickly).

Conceptually, the more customised an aircraft is the more likely it is that residual values are likely to be lower as there will be fewer buyers for them on the secondhand market. Furthermore, previous attempts to stretch the range of aircraft have rendered them significantly less efficient and less suited to deployment elsewhere in the airline’s network.

And finally, the high cost of operating ULH flights requires much higher yields and unit revenues, not just to compensate for the higher costs, but to increase margins. It’s a point of nuance, but Project Sunrise needs to outperform incumbent routes to justify the risk. Simply matching the performance of incumbent routes doesn’t justify the capital expenditure and risk!

Let’s consider these two constraints in more detail. We’ll deal with the fleet question first, and come back to yield, revenue and cost after.

Fleet: the A340-500

We need to take a step back into history and revisit the A340-500. The A340-500 was a progenitor of the current generation of ULH aircraft and its failure has important lessons for contemporary ULH strategies.

It was first introduced in 2002 and is a “shrink” of the A340-600, made shorter by removing several frames from the fuselage, but kept the rest of the structure and wings, and a very similar engine. The resulting aircraft was significantly smaller than the A340-600, making it a similar size to the A340-300, but because it shared the A340-600’s bigger wings, engines and fuel capacity, it lifted a much larger fuel load and carried it over an incredible distance. But there was a catch: while the burned fuel like the A340-600, it carried a similar payload to the A340-300!

It was a niche aircraft and only a handful of carriers ordered it, most notably Thai Airways and Singapore Airlines who were the only carriers operating them on ULH routes. Emirates and Etihad also ordered them, albeit not operating them on ULH routes, instead utilising its lifting power.

In order to operate it on ULH routes, Thai and Singapore both required very low density configurations. Thai introduced it with 215 seat configuration with only 53% of the seats in economy class (C60 W42 Y113). Comparatively, their B747-400s had a 84% economy class density. Singapore went with a straight 100 seat business class configuration.

Thai and Singapore’s low density cabin was not a choice, rather forced on them by payload restrictions required to enable non-stop flights from Bangkok and Singapore to New York. Just shrinking the A340-600 wasn’t enough to enable the non-stop flights and both airlines required a limited payload to make the trip. A secondary consideration was that the premium heavy cabins enabled higher yields and unit revenues to overcome the higher costs (more about that later).

Reinforcing our initial framing, this meant the A340-500 was a niche aircraft. The low density configuration made it difficult to utilise elsewhere in the network, effectively making them orphan fleets. But it wasn’t just the the configuration but also the higher fuel burn on comparative sectors. Let’s put this in perspective:

• We previously noted that the A340-500 was a shrink of the A340-600, giving it a similar size to the A340-300, but with the fuel capacity and lifting power similar to the A340-600. See the scaled image below showing the lengths of the three aircraft remembering that they had the same cross section.

• The implications are profound: while the A340-500 has a similar maximum payload to the A340-300 (54t versus 52t), it has a much higher structural weight with an indicative OEW (operating empty weight) of 168t, 37t higher than the A340-300 (131t). The heavier structure was required to carry the larger fuel load required for ULH flights. For comparison, its maximum fuel capacity is 175t, compared to just 110t on the A340-300.

• This means that the A340-500 is carrying an extra 37t of structure for just 2t more potential payload. Unless you need that extra structure to carry fuel for an ULH flight, you’re carrying 37t of deadweight around your neck!

• Meanwhile, when carrying the extra structural weight on the A340-600 (OEW 174t), you’re also able to carry a much larger potential payload (66t). The extra 12t or 14t of payload makes it more worthwhile as you can generate more revenue to overcome the higher costs! The net result is that the A340-500 is a gas guzzler, and unlike the A340-600 it’s not guzzling the fuel to carry more payload, but simply to fly further.

• Rough estimates put the A340-500 and A340-300’s fuel burn at about 8.0t and 6.5t per hour in cruise, meaning the A340-500 burns 23% more fuel than the A340-300 on a comparable mission. But as we’ve just observed, it doesn’t carrying 23% more potential payload. It burns 23% more fuel to carry just 4% more payload and highlights why it made it cost prohibitive to operate the A340-500 on anything other than ULH sectors.

A few years down the line, Thai (2012) and Singapore (2013) ended their ULH flights and withdrew the A340-500s from service. They were unable to generate the necessary yield and unit revenue premium to make these flights viable. The aircraft were too expensive to utilise elsewhere in their network and, absurdly, it was cheaper to park relatively young aircraft than fly them elsewhere. They were less than 10 year old at the time!

Fleet: enter the A350-900ULR

After a 5 year hiatus, Singapore returned to ULH flights in 2018. It’s doubtful that the market changed that much of the 5 years, i.e. beyond cyclical variations in demand, it’s doubtful that customers were willing or able to pay significantly more for it in 2018 than 2013. Instead, the change has been on the cost side, with the significant improvements that the A350-900ULR offered. Let’s consider the aircraft in more detail.

The A350-900ULR’s long range capability isn’t generated by shrinking the aircraft like the A340-500. Rather, it took the A350-900 and significantly increased its fuel capacity while making a more typical trade of payload for range. This was achieved without larger fuel tanks, but modifications to the aircraft’s fuel system, including structural and software modifications allowing greater utilisation of its existing fuel capacity.

This means that the A350-900ULR doesn’t have the exaggerated fuel burn of the A340-500. In fact its fuel burn is similar to the baseline A350-900. Two A350-900s (one baseline and one ULR) with the same payload and fuel load will burn a nearly identical amount of fuel on the same route.

Let’s look in more detail:

• Singapore’s A350-900ULRs have a MTOW of 280t. If it carries it’s maximum fuel load of 129t its usable payload is just 16t (using an indicative OEW of 135t).

• While the baseline A350-900 has several MTOW options, let’s use the same 280t MTOW version to allow a like-for-like comparison. Its maximum fuel load is just 108t and assuming the same OEW of 135t, its usable payload would be 37t.

• Note, the A350-900 is now available with a higher 283.9t MTOW, however 280t was the highest option when Singapore’s ULRs were manufactured. Presumably, they’d take the higher version now if they were ordering the ULR now. For the sake of the comparison we’ll stick with 280t.

• So the A350-900ULR can carry 21t more fuel, but that costs 21t of payload. It’s a sizeable tradeoff since every 1t of extra fuel the ULR carries compared to the baseline A350-900, 1t of payload is given up. This clearly shows how the A350-900ULR is trading pure payload to carry the fuel, but they’re doing this without compromising fuel burn.

• You may have noticed that there’s an externality to this. In trading off payload to carry the fuel, it’ll significantly limit the number of passengers that the aircraft can carry on ULH missions, far more significantly than the A340-500, but it’ll be a whole lot cheaper and less risky to do it!

The trade-off means that the A350-900ULR will still have to utilise a premium heavy, low density configuration. Singapore’s A350-900ULRs carry just 161 passengers (C67 W94). For comparison, Singapore also operates baseline A350-900s in medium and long haul configurations, carrying 303 (C40 Y263) and 253 (C42 W24 Y187) passengers, respectively.

While the operational and trip costs of the A350-900ULR will be comparable to the baseline A350-900s, they will be constrained by the configuration. This would still limited its ability to be deployed elsewhere across the network, however doing do would only be limited by cabin configuration, and not fuel burn like the A340-500.

Importantly, it has relatively lower residual risk. If the ULH routes fail, the A350-900ULR could be converted into a baseline A350-900 following a cabin reconfiguration and some software update, both things that an airline would do several times during the course of its normal working life. So if there’s a dramatic market shift or the project just doesn’t work, they’re not left with an expensive aircraft corroding away in the desert.

It would be remiss of us not to note the A350-900ULR’s limitations. Even with its much larger usable fuel load, Singapore’s configuration isn’t just a function of their desire to generate higher yields and unit revenues, but also its operational limitations. Its fuel capacity isn’t infinite and 161 passenger configuration is limited by the payload restrictions required to carry the larger fuel load. Simply put, it’s not making the non-stop trip with 200 or more passengers, so reducing density is a necessity, thereby generating a range of externalities including Singapore’s limited ability to generate higher utilisation by rotating aircraft through the network. So you don’t see Singapore sending the A350-900ULR on shorter runs to Bangkok or Jakarta during the day other than IROPS or AOG scenarios.

Fleet: enter the A350-1000ULR?

So if the A350-900 worked for Singapore, then why didn’t Qantas just copy them? Quite simply, it doesn’t carry enough fuel. Sydney-London is 900nm further than Singapore-New York, and additional payload restrictions would’ve been catastrophic! And even if it carried more fuel, what would the remaining payload be? Rough guess is that it would’ve been close to zero!

Qantas needed a new aircraft. While they considered various options from Airbus and Boeing, they chose the A350-1000ULR. We’re not going to analyse the choice of the A350-1000ULR against its competitors. It’s a sunk cost, so there’s limited value from considering the advantages or disadvantage of one or the other.

The A350-1000ULR bears similarities to the A350-900ULR. Airbus have used the same playbook to increase the aircraft’s fuel capacity. Initially it was thought that the increase in usable fuel would be achieved using the same fuel system modifications, however it’s now confirmed that the A350-1000ULR will also sport an additional fuel tank. The RCT (Rear Centre Tank) is a permanent fixture and similar in design to the RCT used on the A321XLR, albeit a lot bigger, carrying an additional 16t of fuel.

When combined with a boost in the MTOW to 322t, the A350-1000ULR’s range will be extended sufficiently without requiring the extremely low density cabin that Singapore requires on their A350-900ULRs. This doesn’t mean Qantas can deploy a similar cabin density to their other long haul aircraft like the A380 and B787-9, rather that it has somewhat more flexibility than Singapore.

Qantas will deploy the A350-1000ULR with a 238 seat configuration with 59% economy class configuration (F6 J52 W40 Y140). This compares to 70% economy class configuration on their A380s and B787-9s (notably, airlines tend to maintain a similar premium mix across their long haul fleets). Comparatively, Singapore’s economy class configuration varies between 70% and 74% on their A380s, B777s and long haul A350s, yet they have no economy class on their A350-900ULR due to operational necessities.

This gives Qantas a more versatile aircraft, while also reducing operational risk and increasing flexibility compared to Singapore as it’ll be easier to deploy it elsewhere across the network . Similar to Singapore, residual risk is reduced as the aircraft can be reconfigured and integrated into the regular fleet if needed. Just a reminder, Qantas have an additional 12 non-ULR A350-1000s on order with additional options to replace the A380s in the next decade. This highlights and reinforces the existing economies of scale and reduced residual risk.

In the same way that we considered the externalities of Singapore’s ULRs, we should consider the same for Qantas’s ULRs. The most significant externality isn’t the cabin density. In fact we’d consider that an advantage compared to Singapore’s strategy. The externality is the higher structural weight of the A350-1000ULR due to the RCT. While it’s not in the same realm as the A340-500, it can’t be ignored. Unfortunately, the specific weight of the RCT hasn’t been published by Airbus yet, limiting our analysis in this regard.

Prices, yields and unit revenue

Is Project Sunrise a bet on whether or not people will pay higher prices for a non-stop flight? It’s a popular narrative, but somewhat of an oversimplification.

Qantas have been open about how they expect to generate a 20 to 30% revenue premium, but this has been misinterpreted as 20 to 30% higher prices. They’re not the same thing! So how do they generate a 20 to 30% revenue premium without increasing prices by 20 to 30%? Higher yields, right?

“Yield” is one of the most misused term amongst avgeeks, mostly because it’s often confused with “unit revenues”. A typical example is when our dear friend Caleb says Airline X should fly so-and-so route because it’s “high yielding” and, as all avgeeks know, high yielding means it’s a sure thing, right? The only more egregious use of terminology is when passenger load factors as used as evidence of a route being successful.

By themselves, yield and load factor are meaningless! You can fill any route, generating a 100% load factor if you sell tickets cheaply enough. Alternatively, you can generate stellar yields with just one passenger. Both routes will fail miserably, so Caleb’s strategy of just flying high yielding routes isn’t going to work, nor will the alternative of maximising load factor. However, a route with high yields and a high load factor will be very successful, because high yields combined with high load factors generates high unit revenues! But what do these terms even mean?

Yield is the revenue earned per passenger, often weighted by distance to allow comparison across routes (i.e. revenue per revenue passenger kilometer). It should now be obvious that yield by itself is problematic, since if you only sell one ticket on the flight but sell it for an incredibly high price, you’ll have a very high yield. And that’s where unit revenue comes in, instead of dividing revenue by passengers, it divides revenue by the available seats on the aircraft (i.e. revenue per available seat kilometer).

Meanwhile, the load factor is simply the number of passengers divided by the number of seats available. A keen mathematical eye (read: high school maths) would notice that unit revenue is the product of the yield and the load factor. All else held constant, maximising unit revenue is what maximises the airline’s revenue on a flight, not yield or load factor by themselves.

When trying to maximise unit revenue, or at least increase it, the airline needs to either increase the yield or the load factor. In the real world, yield and load factor are not independent of each other. Increasing the load factor often involves a trade-off with yield as you will likely have to sell more cheap tickets to increase the load factor, thereby reducing the yield.

But we digress, back to the point at hand. Let’s consider a practical example of how Qantas prices non-stop and one-stop flights using the existing Perth-London flights, and how it is aims to increase unit revenues, not just yields, on Project Sunrise. Perth-London passengers have two options:

1. Taking the non-stop Perth-London flights (QF9/10), or

2. Connecting through Singapore, taking QF71 from Perth to Singapore and then connecting onto QF1 (originating in Sydney); taking QF2 to Singapore and connecting onto QF72 on the return.

Qantas don’t necessarily price tickets higher on the non-stop flight, at least not in economy class. For example, picking a random date in the future we see the non-stop is actually priced slightly cheaper than the one-stop via Singapore. The variance is isn’t due a difference in base fares, but rather the additional airport charges and government taxes for the stop in Singapore. The base fare, i.e. what Qantas is charging, is the same for both the non-stop and one-stop flights.

If you bring the date closer, you start to see a variance between the flights with the non-stop flight now more expensive than the one-stop option. Critics might push back at this point, see it’s more expensive! Not so fast!

The fare on the one-stop option is in a cheaper fare class (i.e. saver fare), while the non-stop is a more expensive fare class (i.e. flex fare). There’s a crucial nuance here: the base fares in each fare class is the same. So the prices are the same, but one or both of the following are happening:

1. The cheaper fares are getting sold more quickly on the non-stop flight, and/or

2. There are fewer cheaper fares available.

The irony of the 1st point isn’t lost on us. How could that be with everyone on social media telling us they don’t want non-stop flights? Well, it’s always the dissenting voices that are the loudest!

The 2nd point should be obvious since there are simply fewer economy class seats available! Let’s consider this with a conceptual example of a very simplified market and assume Qantas offer three economy class fares of $1200, $900 and $500. These are made-up bullshit numbers, but bare with us:

• On the one-stop flight via Singapore, capacity is limited by the smaller A330-200 that flies the Perth-Singapore leg (rather than the A380 that flies the Singapore-London leg). Qantas’s A330-200 has 243 economy class seats. Simplistically, if we assume they sell one-third of the seats at each price point (i.e. 81 seats at each price point), this would generate total revenue of $210,600 and unit revenue of $867 per seat.

• On the non-stop flight, they fly the B787-9 that only has 166 economy class seats. If they followed the same strategy of selling one-third of the seats at each price point (i.e. 55.3 seats at each price point), they would generate a lower total revenue of $143,780, but the same unit revenue of $867 per seat.

• But why would they do that? If they were able to sell 81 seats at $1200 and 81 seats at $900 each on the A330, then why wouldn’t they just sell 81 seats on the B787-9 at $1200, and 81 at $900, and only the remaining 5 seats at $500.

• This is exactly what they would do, and this generates total revenue of $172,600, resulting in unit revenue of $1037 per seat, 20% higher than the A330, and this is the 20% revenue premium!

The smaller economy class cabin means there are simply fewer seats to fill and by filling them from the front they generate a unit revenue! It’s a gross over simplification, but highlights how lower density configurations can generate revenue premiums - in the same cabin - without having to increase ticket prices. As a reminder, Qantas’s A350-1000ULR will only have 140 economy class seats, even fewer than the B787-9, despite it being a larger aircraft!

Premium cabins are different though and we would expect to see higher prices for the non-stop flights. For example, looking at the same dates as the previous example, we observe a higher base fares in business class for the non-stop compared to the one-stop flight for the same fare classes. However, these premiums are nowhere near the 20 to 30% quoted earlier, rather about 7% and 9% for business saver and flex, respectively. Premium economy is unavailable on the A330, so there is no point of comparison, while first class isn’t available on either aircraft.

Even in the premium cabins, the revenue premium isn’t generated by higher prices alone, but by smaller cabins. Comparatively, the A350-1000ULR will have 6 first class seats compared to 14 on the A380 that currently flights Sydney-London, 52 business class compared to 80 on the A380, and 40 premium economy compared to 60 on the A380, and 140 economy class compared to 341 on the A380. There’s a lot of scope to increase unit revenue by filling from the front!

Higher unit revenues, but won’t costs also be higher?

In our analysis of the A350-900 and A350-900ULR we noted that the cost of operating them on the same (shorter) route would be similar. Specifically, we highlighted how carrying the same payload over the same distance would burn a similar amount of fuel. We can extend this to the A350-1000 and A350-1000ULR, albeit the ULR will be slightly more costly since it’ll be carrying the additional structural weight of the RCT fuel tank, however we don’t expect this to be very large.

This is a distortion view of the cost dynamics since it only considers "trip cost”. While the trip cost will be similar, the unit cost or average cost per available seat kilometer will be different. Since the A350-1000ULR will seat fewer passengers than the A350-1000, it’ll have a higher than the A350-1000, with the specific magnitude dependent on the configurations of the aircraft.

We don’t know the configuration of Qantas’s non-ULR A350-1000s yet, but it’ll certainly be less dense and have more seats than the 238 seat ULRs. For comparison, British Airways, Cathay Pacific and Qatar Airways seat 331, 334, and 327 on their A350-1000s, respectively. If Qantas follow suit, this would mean that their fuel unit cost of the ULR would be approximately 39% higher than the baseline A350-1000, at least on the same route.

But fuel isn’t the only cost meaning that this cost premium isn’t directly comparable to the revenue premium. In FY25, fuel accounted for 24% of Qantas’s expenditure. If we simply scale this it’ll translate into a 9% unit cost premium, still well below the revenue premium of 20% to 30% that they are targeting, generating a significant boost in margins.

Even then, a 9% higher unit cost might be an overestimate as non-stop flights will generate other cost savings, particularly lower crew costs (single crew working a longer flight is cheaper than two crews working two shorter flights, especially when taking into account the extra stopover), lower airport costs (see the example of the Perth-London non-stop), IROPS costs from additional connection, etc.

While we don’t have a lot of hard evidence to go on here, we can’t simply isolate the cost impact to reduced cabin density. But even if we did, we must consider that Qantas haven’t had to compromise cabin density to the same extent that Singapore have, reinforcing our earlier discussion.

Yet, we’ve also seen Qantas show a willingness to compromise unit costs by employing payload restrictions to enable some of its current ULH routes. Dallas-Melbourne is a good example of this: Qantas operate 4x/week flights, increasing to daily during peak season, operated by the B787-9. This is Qantas’s 2nd longest non-stop flight, however the westbound leg encounters significant payload restrictions due to en route weather. We’re able to investigate this more utilising USDOT data.

Over the last 12 months, Qantas’s declared capacity on the eastbound leg averaged 235 seats, compared to just 225 on the westbound leg (the B787-9 seats 236 passengers). This indicates an average gap of 10 seats, meaning the payload restriction on the westbound leg averaged 10 seats more than the eastbound leg over the last year. For comparison, there was no gap between the east and westbound legs of their Melbourne-Los Angeles B787-9 flights over the last year (excluding those operated by the A380).

There’s a lot of seasonal variation in westbound payload restrictions. There was no gap in October, while the gap remained less than 10 between July and January, however it grew from February to a peak of 22 in April.

The impact of payload restrictions is profound since it reduces the denominator that costs are shared over. An average payload restriction of 10 seats implies a 4% higher unit costs. While this may appear trivial, it highlights that Qantas are willing to tolerance higher unit costs in search of higher yields and unit revenues on ULH flights.

Given Qantas’s recent capacity increases on the route, it is a tacit acknowledgement of the revenue premiums they’re generating from ULH flying. It goes a long way to reinforce their confidence that such premiums are exceeding the marginal cost sufficiently.

But there’s a caveat: on ULH missions, one carries fuel to carry fuel! Simply put, an A350-1000 taking off at 283t is going to burn more fuel over the same route as the same aircraft taking off at a lower weight. Thus, a 22 hour non-stop flight is going to burn, on average, more fuel per hour to carry the same payload than the same aircraft and payload taking a well-timed fuel stop halfway.

But here’s where the A350-900ULR and A350-1000ULR have taken a giant leap forward. Unlike the A340-500, they’re not carrying the same deadweight! As fuel is burned during the flight the aircraft gets lighter, however the A340-500’s excessive structural weight to carry the huge fuel load just didn’t get lighter as the flight wore on.

While this may appear abstract, the fact that the A350-900ULR and A350-1000ULR aren’t meaningfully heavier structures compared to their baseline peers means that they’re more efficient than a previous generation of ULH aircraft. The “delta” to carry the extra fuel for their ULH mission is significantly less than their predecessors.

Project Sunrise isn’t that original!

As our discussion has highlighted, Singapore Airlines have been pushing the boundaries of ULH flying for some time, while Qantas have been doing the same. Project Sunrise will pushing the boundaries further, and significantly so with Sydney-London being 900nm or 11% further than Singapore-New York. Yet, our thesis speaks of it as somewhat inevitable, driven by the technological gains and Australia’s “tyranny of distance”.

While opposing voices will be the loudest, we might look back on this in a few decades as the inevitable evolution from 5-stops to non-stop. And if you don’t like it, then don’t do it. In addition to the plethora of one-stop connecting options, Qantas have indicated that they’ll also maintain one-stop flights to London (via Singapore) and New York (via Auckland).

Many commentators continue to argue that it’s just PR and spin. Yes, Qantas are very, very good at this, but one can’t ignore the technological, regulatory and operational challenges that Qantas have had to, and still need to overcome. They’ve had to order a highly customised aircraft, engage with regulators to reform flight duty limits, and renegotiate employment terms. Nevermind the massive reorganisation of their network that will have to occur (e.g. have a look at our previous analysis on the difficulty in coordinating their existing slot portfolio at Heathrow simply to schedule the Sydney-London flight). This seems like a lot of work, money and risk for a stunt!

So if it’s not about the PR, then what is it?

So far we’ve looked at the technological, operational and economic considerations, but it hasn’t answered the question why they feel compelled to take the risk. These are actually more strategic in nature.

First and foremost, it’s about competition. Qantas’s one-stop flights are forced to compete on price and/or product with other airlines, and in many cases, they’re at a significant cost disadvantage, making it a battle they simply can’t win. This has been a significant contributor to the reduction in their London capacity over the last two decades.

When it comes to Project Sunrise, the competition can’t or won’t do it! This may appear flippant, but it’s at the core of the strategy. Simply put, there are no other airlines that have a fleet capable of non-stop flights between Melbourne or Sydney and London, nor are they likely to.

British Airways are the only other airline with direct flights between Australia and London. They don’t have an aircraft capable of making the non-stop trip and Qantas are making a relatively safe bet that BA won’t suddenly order the A350-1000ULR. Firstly, it simply doesn’t make sense for BA to acquire the niche aircraft on just one route. BA won’t have the scope to build scale over several routes and there aren’t many other 22 hour routes from London that BA would want to fly that would enable them to build a fleet of A350-1000ULR aircraft. Secondly, even if they were to do so, Qantas have the first mover advantage, reducing the likelihood of BA doing it.

The same factors also apply to Virgin Atlantic (or Delta UK as some might tease), who have shown little interest in recent years beyond Trans Atlantic flights or flights that benefit from Joint Venture feed. The other Virgin, Virgin Australia are only operating long haul flights on wetleased aircraft at present. They are a long way away from returning to their own long haul operations, nevermind trying to operate a niche sub-fleet of ULH aircraft. It could take a long time before Virgin could even consider it, and more to operate it if they wanted to. It would appear as thought Project Sunrise would give Qantas a monopoly on non-stop flights to London, and probably to Western Europe as well.

A similar case can be made to the US. Are United, Delta or American Airlines going to acquire a special fleet simply to operate non-stop flights to Australia? Again, the question would be are there other ULH routes that may interest them to build scale around an ultra long haul business? The scope is a little broader for the US carriers, but despite their growing international ambitions, there have been no suggestions that they’re interested in it. US carriers haven’t attempted to cover Singapore’s New York routes, and there’s no indication that they’d do the same to cover Qantas’s New York route.

The lack of (potential) competition on non-stop routes gives Qantas an advantage that competitors can’t match, or at least are unlikely to. In fact, Qantas’s most significant competitors to London aren’t BA or Virgin. It’s the one-stop options, primarily Emirates, Qatar, Singapore, Cathay, etc.

The most recent market share data between the Australia and UK shows Qantas and BA with just 20% and 5% of the market, respectively. The remaining 75% is held by one-stop carriers, led by Emirates, Qatar and Singapore with 23%, 17% and 12%, respectively. It’s this 75% that Qantas are targeting with near certainty that they’ll have a monopoly on non-stop flights between Australia and the UK!

In addition to the strategic advantage, the lack of competition brings two practical advantages. Firstly, pricing power, allowing Qantas to generate higher yields and unit revenue, particularly from the front cabins, but secondly reducing the need to discount to attract or defend market share!

At the same time, Qantas are being rewarded for their patience. While Singapore got to market first, Qantas’s patience means they entering the stage with a better solution and considerably lower risk. A dalliance with the the A340-500 would’ve cost Qantas a lot of money and set them back a long time. The evolution to Project Sunrise isn’t a zero risk strategy, however they’re not facing a catastrophic scenario if things don’t go as planned. The aircraft will simply be converted into a regular configuration and fly on across the rest of the network.

But really, why are they doing it?

By this point it’s taken a lot of convincing and not everyone would be convinced by this argument. Despite the relatively lower risk compared to a decade or two ago, it’s still a risky endeavour. We’re made the operational and strategic case, but it still involves a leap of faith. These arguments don’t explain why Qantas feel they need to take the risk. They’ve have had a run of incredibly strong results, so why now?

We titled the article “how to think, not feel about Project Sunrise”, yet the question we pose now asks “Why does Qantas FEEL compelled to take the risk?”. We’re on shaky ground here, so one final point of analysis to nail the argument:

Qantas’s international operations are the worst performing part of the Qantas Group. This isn’t a shot at Qantas International and shouldn’t be misinterpreted as an argument that they’re performing poorly, rather it’s an empirical point that Qantas International is lot less profitable than Qantas Domestic and Jetstar. This relative profitability within the group has implications for capital allocation.

In FY25, Qantas International contributed just 25% of airline EBIT (earnings before interest and tax), despite accounting for 41% of airline revenue and capacity (measured by available seat kilometers). We’re referring to airline EBIT and revenue, so stripping out the non-airline parts of the business including Qantas Frequent Flyer and corporate.

This isn’t a recent trend and Qantas International didn’t suddenly have a poor year. In fact, FY25 was a very good performance, as the years leading up to the pandemic saw its share of airline EBIT consistently lagging its share of airline revenue and capacity, and by even wider margins in most years.

Some commentators argue that Qantas International’s profitability is undervalued given its importance to Qantas Frequent Flyer (QFF). QFF is an immensely profitable part of the group, earning A$ 556 million in EBIT during the last financial year. Let’s humour this for a moment: if we generously add all of QFF’s EBIT to Qantas International, it would contribute 39% of EBIT, broadly in line with its share of capacity (41%). However since QFF’s revenue is also now included, the share of EBIT is still lower than the share of revenue that now stands at 47%. FY25 is consistent with historical trends, including the periods before the pandemic.

Thus, it would require the most generous assumption for Qantas International’s performance to even be comparable to the rest of the group, highlighting a clear underperformance of Qantas International. The reason for this is fairly simple: Qantas International faces a lot more competition compared to the Qantas Domestic as the domestic market faces larger barriers to entry. While this is somewhat beyond the scope of our analysis, the implications are relevant.

At a group level, it’s not rational to disproportionately deploy capacity - and thus capital - towards the least profitable part of the group. In the long run this has the potential to drag the rest of the group down with it. This is particularly consequential for an airline given the highly capital intensive nature of the business. Again, this isn’t the focus of our current analysis, but have a look at our previous work on the topic to give more context why this is so important:

As noted, Qantas International’s relative underperformance isn’t a new phenomenon, and it’s arguably already had consequences for the group’s capacity allocations. Qantas International accounts for 41% of ASKs, down from 46% in FY19. However, if we go back even further back, Qantas International accounted for 77% of group ASKs in FY01 (it’s the furthest we could quickly find). It’s a different baseline (i.e. no Jetstar), however it clearly shows how Qantas have steadily “underinvested” in Qantas International’s capacity.

This shouldn’t be interpreted to mean that they’ve marginalised the business, rather the more profitable parts of the business have attracted capacity more rapidly. And before anyone quips that Qantas International’s poor performance is due to this underinvestment, let’s remind readers that we’re not using the term in an absolute manner, rather relative. In fact, Qantas’s mainline fleet is slightly younger (average age of the A330, A380 and B787 fleet is 15 years) than the domestic mainline fleet (average age of the B737 and A321 fleet is 17 years).

My head is spinning right now, so can you pull it all together now?

So here is our hypothesis: Qantas is in the early stages of an extraordinary and near total fleet renewal. With the exception of the B787s and a handful of newly delivered A321XLRs, every mainline aircraft will be replaced in the next decade. The order book already includes replacements of all B737s, A330s and A380s, with the bulk of the B737s and A330s being replaced within the next 6 years.

The relative performance of Qantas International makes the business case for this investment relatively weak compared to Qantas Domestic and Jetstar. To justify this incredible capital expenditure, Qantas International has to make a deal with the devil by increasing its margins, and by a lot! If it’s unable to do so, the board will have little option but to continue the relative underinvestment in Qantas International, limiting their capacity growth or even shrinking capacity.

Arguably, the A350-1000ULRs have been ordered to provide incremental capacity growth as they’re not replacing an existing fleet, at least for the time being! The rest of the order book is replacement capacity: the A330s will be replaced by a confirmed order for more B787s and A350-1000s (non-ULR version). However, it’s notable that the A380s, which are due for replacement from 2032 or thereabouts, don’t have confirmed replacement orders. While Qantas have confirmed that the A380s will be replaced with more A350-1000s, these orders aren’t confirmed (at least based on most recent AFS) and will utilise purchase options that were part of the previous orders.

And this is Qantas’s failsafe: if Project Sunrise doesn’t live up to expectations and if Qantas International are unable to increase margins significantly between now and 2032, or at least show some evidence of their ability to do so, the Board won’t hear a convincing case to exercise the options to replace the A380s. Instead, it’ll make more sense to limit Qantas International’s growth, or maybe even reduce capacity or shift capacity to Jetstar or joint venture partners, all things they’ve done in the recent past. Instead, the A350-1000ULRs could/would become the de facto A380 replacement.

In the introduction, we pointed out that Cam Wallace, the CEO of Qantas International, initially wrote-off Project Sunrise as a publicity stunt. So we wondered what changed Cam’s mind? We think Cam was convinced that Qantas International needed a paradigm shift to increase margins and maintain its relevance within the group. Gimmicks and small changes weren’t going to drive the structural shift needed in margins.

Project Sunrise is a paradigm shift! But it’s also a bet, a big bet, that they can increase margins by shifting a significant portion of capacity to routes with limited competition, allowing them to gain pricing power and exploit lower density cabins to increase unit revenues (doing so significantly quicker than unit costs will increase). Moreover, it does this with a relatively low risk failsafe! While it’s not zero risk, it’s not taking on same degree of risk that Singapore and Thai did with the A340-500.

Now before you think it’s too fanciful, we’re not trying to suggest that this was a grand strategy plotted in a smoke-filled room more than a decade ago. Instead, Qantas International faced a challenge: develop a strategy to increase margins to justify their capacity! They were simply sick and tired of seeing the relative investment in Qantas International declining over time.

The problem that lead them down the path is there aren’t many things that can systematically increase their margins. They’re already exceptionally good at generating strong yields and competition limits their ability to increase this further without increasing costs quicker. Frankly, they’re already a pretty efficient operation. It’s not like there’s a trick they’re missing on the cost side. No doubt, everyone has their pet idea, something that they think can improve margins, but Qantas recognised the need for a paradigm shift, rather than small ideas.

While the idea isn’t novel, the relative success of pushing the envelope with the B787 on ULH routes like Perth-London, Melbourne-Dallas and Auckland-New York, has convinced them that the concept works. They’ve seen the data and it shows them the secret sauce: increasing unit revenues quicker than costs increase, thereby increasing margins. It’s reinforced their confidence that this will also be the case from Sydney and Melbourne, non-stop to London and New York!

It was just a matter of when to roll the dice. They remained patient, waiting for the technology to find the sweet spot where they could do this without risking the whole show! They’ve eventually found their aircraft and pulled the trigger!

Hopefully this provoked you more about how to think, and not feel about Project Sunrise! Thanks for reading, and please share it with colleagues and friends!

Share

This short article outlines a technical thread we posted on social media on Sunday morning, highlighting Qantas’s IROPS masterclass! We thought we’d repost it here with a little bit more detail to work through the IROPS challenges and implications, and quite frankly, just how well they responded!

Additionally, our post was plagiarised, word-for-word, by at least one publication. We wanted to post it here to ensure our analysis was documented. As we’ve noted before, it’s why we’re considering placing our work behind a paywall and promoting the use of an ethical paywall.

Background

Thunderstorms at SYD and CBR meant QF22 DFW-MEL had to take unscheduled fuel stop at AKL, while QF12 LAX-SYD and QF8 DFW-SYD both diverted to BNE. Why does SYD and CBR weather force MEL bound flight to stop in AKL for fuel you ask?

Poor weather was forecast on Sunday morning at Brisbane, Melbourne and Sydney, as highlighted in the daily ATFM plan published by Air Services Australia. The forecast wasn’t catastrophic, however it included the chance of thunderstorms at both Brisbane and Sydney. A particular challenge with thunderstorms is their localised impact, while their exact timing and location can be difficult to predict. This makes planning flights around thunderstorms difficult with a lot of uncertainty.

QF22 DFW-MEL

• QF22 DFW-MEL, operated by a B787-9, made an unscheduled fuel stop at AKL. It seems obtuse that thunderstorms at SYD and CBR meant a flight to MEL needed to stop along the way for fuel, however QF22 couldn’t use SYD or CBR as their alternate since poor weather was forecast. For flight planning purposes, one needs good weather at your alternate or a 2nd alternate further away with better weather.

• This meant that QF22 needed to carry extra fuel for an alternate further away, however the flight is already carrying a significant payload penalty and pushing physical fuel carrying capacity. It’s one of the longest flights in the world and already operating at the edge of the aircraft’s capability.

• It’s likely that QF22 simply couldn’t carry any more fuel, or would’ve required a catastrophic payload penalty to carry fuel the additional fuel required. Instead, QF22 was replanned to depart DFW for AKL. The ADS-B data clearly shows the aircraft heading directly for AKL after departure.

• Taking a tech-stop at AKL generates several externalities as crew will exceed their duty limits and time out at AKL. It’s not like airlines just keep B787 crew hanging out at AKL just in case. While Qantas do operate B787 flights to/from AKL, it’s not a B787 crew base. This meant that an alternative crew had to reposition to AKL. This may have occurred ahead or time, or possibly the crew was switched with QF3/4 (SYD-AKL-JFK flight).

• ACARS messages show significant care to mitigate passenger disruptions. For example, this included arranging catering for additional leg in the middle of the night and rebooking of connections. We really loved this one:

• Onward business class passengers from MEL to PER were given option of rebooking on immediate onward connection without their bags, or a later connection with bags, however this would be in economy class due to limited seats available on that flight. Furthermore, the message gives a clear indication that ground staff were prepped to help, but also giving passengers autonomy to make decision!

• All in, they arrived at MEL just short of 3 hours late which is good going for unplanned stop which required crew change! But that’s not all …

QF12 LAX-SYD

• QF12 LAX-SYD, operated by an A380, diverted to BNE. Unlike QF22, this decision wasn’t planned ahead and a relative late decision was made a few hours out of SYD. This is obvious from the ADS-B data which shows the flight clearly heading to SYD.

• Weather at SYD was somewhat worse than expected with the thunderstorm expected over the airfield near landing time which would’ve required significant additional holding. Waiting that out at SYD would’ve risked landing below minimum fuel or encountering a fuel emergency.
  

• The diversion to BNE was ostensibly for fuel, however the much shorter length of the flight compared to QF22 meant that if the aircraft were refuelled quickly they could continue to SYD without exceeding the crew’s duty limits.

• No crew had been repositioned and there was a significant risk of hitting duty limits. If the aircraft wasn’t refuelled quickly it would’ve required termination of the flight at BNE.

  

• Once again, ACARS messages show s sense of urgency with ground crew prepped to hustle for 45 min splash and dash. That’s clearly aspirational since it was very unlikely to turn an A380 that quickly. Nevertheless, they managed to get out of BNE in time and made it to SYD a little over three hours late.

QF8 DFW-SYD

• QF8 ran into similar troubles to QF12, with an even later decision to divert to BNE. Also operated by an A380, QF8 has a longer block time than QF12 meaning that it was inevitable that the crew would exceed their duty limits as a result of the diversion.

• As a result, the flight was terminated at BNE and passengers rerouted to their final destination on domestic flights from BNE. The aircraft ferried empty to SYD later that day.

• BNE isn’t an A380 crew base and with no relief crew available at BNE it highlights how the proactive decision to replan QF22 to AKL limited the impact of diverting like QF8 did.

Concluding thoughts

In practice, managing IROPS is incredibly challenging. While an individual passenger is focused on their flight, staff at Qantas’s integrated operations centre are handling multiple flights simultaneously. Three flights took unscheduled stops or diverted, but the same attention would’ve had to be applied to countless other flights that didn’t stop or divert.

Furthermore, the environment is dynamic with parameters changing constantly as evidenced by the situation devolving through the night while the flights were already in the air. This is evident by the examples, highlighting how different flights encountered different constraints (e.g. flight duty limits affected each flight differently).

It’s all good having great IROPS plans and procedures in place, but implementing them in a dynamic and evolving environment is difficult, especially when resources are limited. Airlines simply can’t have unlimited redundancy (e.g. it was possible to plan for crew redundancies on the B787 at AKL, but not for the A380s at BNE).

No doubt a lot of passengers will be irate, but this is as good as it gets! There’s no better outcome here, just much poorer outcomes. The public doesn’t get to see what happens behind the scenes. They don’t see how decisions - invariably focused on the least worst outcome rather than an optimal outcome - get made under pressure with imperfect information.

But this isn’t the public’s fault. They aren’t technical experts and we shouldn’t expect them to see the big picture. They will inevitably focus on their flight and the impact on their plans!

We can’t ignore that airlines don’t do a great job at communicating just how complex and difficult this is! Actually, they do a terrible job often dumbing things down and being incredibly vague, and Qantas are far from the worst!

As we’ve argued before when analysing the potential impact of flight delay and compensation schemes, it often appears to the customer that the airline is hiding information or not being truthful.

Airlines are simply too scared to say too much. A culture of risk aversion dictates that they mustn’t say anything that the public may use to hold them to account. So instead they’ll say vague things like “your flight is delayed by weather”. It’s true, but is infuriating to hear because it’s obvious.

Why don’t more they communicate more transparently? Do they think people don’t want to know, don’t care or can’t understand? We’re not suggesting a sinister motive, rather just risk aversion, but we argue that transparency builds trust and shows customers that they’re not the victim of a random bureaucratic decision.

This was a great opportunity for Qantas to communicate the challenges that running an airline encounters and just how professional their handling of it was. It was a masterclass!

Our mate Caleb reckons Qantas hates Melbourne because they no longer have non-stop Melbourne-San Francisco flights despite United still flying the route and because Qantas fly to San Francisco from Sydney. San Francisco being United’s largest west coast connecting hub is irrelevant to Caleb, while Qantas’s strategy of shifting US capacity to Dallas is #fakenews!

On a more serious note, many will argue that Qantas’s cancellation of direct Melbourne-London flight via Perth in July 2024 is evidence of Qantas’s anti-Melbourne bias. Furthermore, as recently as 2012 they operated twice daily direct flights between Melbourne and London.

None of these flights were non-stop, rather direct flights operating via Dubai, Perth, Hong Kong and/or Singapore. The loss of the direct flight may be temporary as its likely to return as a non-stop flight as part of Project Sunrise in the coming years.

The counterargument is that the loss of direct flights to London on Qantas is purely symbolic. Other than a change of flight number, is there really much difference flying on QF37 to Singapore and connecting on QF1 to London (and QF2/36 on the return)? In fact, the outbound to London has a shorter stopover in Singapore compared to the direct flight from Sydney (and identical on the return).

QF1  SYD SIN  3:55pm  9:05pm
QF1  SIN LHR 11:05pm  6:15am⁺¹
QF2  LHR SIN  8:10pm  5:40pm⁺¹
QF2  SIN SYD  7:40pm  6:35am⁺¹

QF37 MEL SIN  4:55pm 10:05pm
QF1  SIN LHR 11:05pm  6:15am⁺¹
QF2  LHR SIN  8:10pm  5:40pm⁺¹
QF36 SIN MEL  7:40pm  6:15am⁺¹

Is it just semantics?

This is a long chapeau, but what it highlights is that these debates are subjective. People will pick the side of the argument based on their personal preferences. So if you think the Perth connection is better than Singapore then you’ll feel aggrieved.

We wanted to weigh in since the topic has come up again and again on social media and we wanted to weigh in while trying to stay clear of subjective preferences. We have our own, and they’re meaningless to anyone other than ourselves. And they’re particularly unimportant to someone who thinks or feels differently.

Instead, we wanted to delve into some data to consider it more objectively. Working through the data we’ve found a few angles to inform the debate. It doesn’t solve it, but rather provides food for thought. The analysis focuses on two different metrics, Qantas international seat capacity over time from various Australian cities, both in absolute terms and relative to Qantas’s total capacity. And secondly, Qantas’s seat capacity over time as a share of total international seat capacity from that city.

Qantas international seat capacity

The first metric to look at is Qantas’s international seat capacity from different airports over time. We’ve limited our analysis to the big four: Brisbane, Melbourne, Perth and Sydney, as Qantas’s capacity from other airports is limited. We only consider outbound capacity as a proxy of total.

One methodological quirk we utilise is that we allow seat capacity to be double counted for flights with more than one stop in Australia (e.g. Qantas’s Sydney-Perth-Paris flights). This might appear counterintuitive since the same seat can’t be sold to two different people. Once a seat is sold Sydney-Paris, it can’t be sold Perth-Paris. However, since we’re looking at capacity rather than utilisation, the seat was available for sale from either Sydney or Perth, so we include it in both cities as a measure of Qantas’s capacity from that city. So what does the data say?

Firstly, it’s clear that Qantas is incredibly Sydney-centric! Sydney accounted for 58% of Qantas’s total international seats in 2024, with Melbourne and Brisbane way back with 23% and 14% respectively, while Perth has just 4%.

Secondly, Sydney’s dominance has waned over time, falling from a peak of 62% in 2014, but higher than the 55% when our data begins in 2006 (length of the time series is limited by BITRE data availability). Melbourne’s relative peak was 25% in 2019, having increased from 22% in 2006, while Brisbane’s current level of 14% equals its pre-pandemic peak (2016 through 2019; 19% in 2021shouldn’t be considered due to the vagaries of pandemic capacity). Meanwhile, Perth’s peak was 7% way back in 2007 and 2008.

These shares only tell part of the story. Qantas’s total seat capacity has ebbed and flowed over time, so a higher share could be a higher share of a smaller pot. Absolute seat numbers are important and relevant:

Since 2006, Melbourne and Sydney’s seat capacity has declined by 8% and 7%, respectively, while Brisbane has surged 45%. Perth has sunk 41%, albeit from a much smaller base. All three cities that saw declines have a similar U-shaped function which has shows some improvement in the lead-up to and in the aftermath of the pandemic (okay, maybe not Perth). Brisbane doesn’t and its growth has been consistent. In absolute terms, Sydney shed 226k seats between 2006 and 2024, far more than Melbourne’s 101k and Perth’s 148k.

There’s no doubt that Qantas is incredibly Sydney-centric, but the data also suggests they’ve become less Sydney-centric over time, but not to a degree that suggests a strong shift in strategy. But there’s something missing from this: Jetstar.

Jetstar is a full owned subsidiary of Qantas and macro-level capacity decisions are made at a group level. Capacity is scarce meaning that Qantas and Jetstar are competing within the group for allocation of capital, including investment in new aircraft. This doesn’t mean they’re competing with each other on an operational level, however the capacity of Qantas can’t be viewed in isolation of the rest of the group.

Something, something, Jetstar

Jetstar are significantly less Sydney-centric than Qantas, and have become even less so over time. Sydney accounted for just 24% of Jetstar’s total international seats in 2024, compared to 58% at Qantas. Melbourne is Jetstar’s leading city, with 29% of their total international capacity, while Brisbane and Perth accounted for 16% and 9%, respectively. Encouragingly, other cities accounted for 21% of Jetstar’s capacity, breaking from Qantas’s strategy of concentrating its capacity in the major cities.

It’s more challenging to consider the trends over time as Jetstar were in its infancy in 2006. In fact, some of the 2006 capacity in these figures includes Australian Airlines, a former Qantas subsidiary that Jetstar’s international operations de facto replaced. Considering a shorter period over the last decade shows exceptional growth at Brisbane, moderate growth Melbourne, and declining capacity shares at Perth and Sydney. Other cities is somewhat more volatile, particularly over the pandemic period. Brisbane’s growth since the pandemic is notable, albeit some of which is due to shifting of capacity from Gold Coast.

Absolute seat capacity highlights Jetstar’s growth story. Despite declining capacity shares, seat levels levels have actually increased in Sydney, but by much smaller magnitudes than at Brisbane and Melbourne.

We’ve previously described Jetstar’s growth in a short analysis in February, highlighting how it’s led capacity growth within the group. In 2024, Jetstar supplied 37% of group level seat capacity, making it somewhat obtuse to consider Qantas in isolation of Jetstar. When combining Qantas and Jetstar capacity, the picture shifts considerably.

Qantas Group

At a group level, Sydney accounted for 43% of total international capacity in 2024. Furthermore, it’s share has declined consistently from 52% in 2006. This decline is rather more significant than when considering Qantas’s mainline capacity in isolation.

Meanwhile, Melbourne accounts for 24% of total capacity, having increased from 22% in 2006, but lower than its peak of 27% prior to the pandemic. Brisbane has also increased significantly from 9% to 16% between 2006 and 2024, while Perth has maintained a consistent 6%. Other cities have declined slightly.

The trends are reinforced when considering data in absolute terms. Brisbane and Melbourne have added 618k and 633k seats each since 2006, or 114% and 48% in relative terms. These increases are nearly double that of Sydney’s 306k or 10% increase.

Viewed from an incremental perspective, i.e. of the additional capacity added by the group as a whole, Brisbane and Melbourne account for 38% and 39% of the net incremental capacity growth, compared to just 6% and 19% at Perth and Sydney, respectively.

Viewed together, these data and trends reinforce the view that Qantas is incredibly Sydney-centric, however it’s become significantly less Sydney-centric over time.

But what about other carriers?

The previous analysis is just one way of looking at the world. It’s comparing Qantas to itself. As the introduction noted, the concern isn’t just Qantas maintaining routes from Sydney while cutting routes from other cities, but that foreign carriers continue to maintain and even grow routes from Brisbane, Melbourne, Perth, etc.

The narrative goes that we can see that Qantas are neglecting Melbourne because of the manner in which foreign carriers pump capacity into Melbourne. For example, Singapore Airlines have 5x/day flights and Cathay Pacific have 2-3x/day flights, yet Qantas only have 2x/day and 1x/day to Singapore and Hong Kong, respectively.

This obviously ignores the network effects that these carriers have at their connecting hubs, but it does raise an important point of comparison. In addition to comparing Qantas to itself, we must also compare their capacity to all other carriers. The point of comparison is Qantas’s international seat capacity as a share of total capacity from each city, rather than Qantas’s total Australian seat capacity.

When considering Qantas in isolation we see that it accounts for 24% of all international seats from Sydney, significantly higher than the 19% and 15% they hold at Brisbane and Melbourne, respectively. Again, Qantas looks particularly Sydney-centric.

One again, it’s also become significantly less Sydney-centric over time, having seen their share decline from 31% to 24% since 2006. Melbourne’s share has also declined from 27% to 15%, while Perth has experienced an even more most rapid decline, from 25% to 7%. Only Brisbane has fared better, seeing their share increase from 16% to 19%.

This analysis undermines the narrative that they’ve become significantly less Sydney-centric over time given the equally significant declined at Melbourne and Perth, however we might also argue that it’s become far more Brisbane focussed. But let’s not forget Jetstar again. The Jetstar story looks entirely different, with their seat share increasing considerably from all cities. Jetstar now hold 12% at Melbourne and Perth, and 13% at Brisbane, while their seat share at Sydney is just 7%.

At a group level, the trends look far less less volatile, and reinforcing the large capacity pivots that we observed in the previous analysis. When combined, Qantas and Jetstar hold 31% of the total seat share at Sydney, having declined from 33% in 2006. However, the share has been consistent for a number of years.

Melbourne’s share has declined from 30% to 27% since in 2006, however it’s shown consistent improvement since a low of 21% in 2010. Similar trends are observed in Perth. Once again, the growth story is Brisbane, with the group holding a 32% share of international seats, having increased from just 19% in 2006.

So does Qantas really hate Melbourne?

The question was facetious, however it’s an important theme that’s helped us navigate the analysis. Instead, the analysis considers the Sydney-centric nature of Qantas and its evolution over time. The conclusion is clear: yes, Qantas is incredibly Sydney-centric, however this is a very narrow view. We highlighted how one can’t look at Qantas in isolation of Jetstar given the important and growing contribution to the groups’s capacity.

Jetstar have grown to account for 37% of group-level capacity. This is even more important when considering how Jetstar has effectively accounted for all group-level incremental capacity growth in the last two decades. Put simply: as a group, they’ve added 1.9 million more seats per year compared to 2006, yet Jetstar have added 2.6 million! When considering group-level capacity, they’re meaningfully less Sydney-centric.

The status chasing frequent flyer among us might ridicule Jetstar’s international capacity. It’s just a budget carrier without a premium product (their business class isn’t business class and should rather be called premium economy). Status credit and points earning isn’t sufficient to get us to platinum status and that first class lounge. But nobody is forcing you to fly Jetstar and even if you don’t, its yield segmentation allows Qantas to increase the number of premium seats on mainline aircraft and generates important price competition in the market. The status chasers among us benefit without even flying Jetstar!

It also shouldn’t be surprising that Qantas is Sydney-centric. It’s not unusual for legacy and network carriers to have a dominant hub. We find similar strategies pursued by legacy and network carriers around the world (e.g. Korean Airlines in Seoul, Japan Airlines and ANA in Tokyo, BA in London, Iberia in Madrid, Air France in Paris). In fact, we might argue that Qantas is far less Sydney-centric than any of these carriers are around their dominant hubs! Exceptions to this are typically the result of unique circumstances, e.g. not every country is as large as the United States that allows multiple airlines to have multiple hubs, and not every country is as spread out as Canada where Air Canada has multiple hubs.

The more surprising result is how Qantas has become less Sydney-centric over time, and not just at a group-level. At a group-level, its held its own in Melbourne, maybe contrary to public opinion that often conveniently ignores the impact of Jetstar. But the rise of Brisbane highlights how the group has sought opportunities to shift capacity growth outside of Sydney. The magnitude of the trend is too large to be an accidental artefact in the data.

This analysis is only as strong as the data we’ve utilised and it has its limits. One caveat is that it only considers seats and doesn’t weight for distance (i.e. available seat kilometers (ASKs)). That’s a more accurate measure of capacity, however the framing of this analysis is somewhat narrow and seat counts are able to highlight the macro-level trends.

More importantly, the analysis isn’t able to account for the most recent trends and future capacity shifts. For example, it doesn’t yet fully account for Qantas and Jetstar’s renewed interest in Perth with several new recent new routes, and several former routes (e.g. Auckland and Johannesburg) returning next month.

No doubt, some readers will disagree with our conclusions, however the data are the data. And that data shows a lot of nuance. If you really believe they hate Melbourne, then maybe the final set of figures would force you to say they hate Sydney too. They certainly don’t, but it reminds us that too much of the hate is just “vibes“!

We hate speculation and we really hate speculation surrounding new routes. Many (most) rumours are simply implausible but it’s difficult to convey this to our Armchair CEO Caleb who heard about the new route from his mate down the pub who was told about it by his uncle’s friend who was a pilot for Ansett. As Brandolini’s law teaches us: “The amount of energy needed to refute bullshit is an order of magnitude bigger than that needed to produce it.”

We prefer to follow the data rather than rumours, speculation and 3rd hand accounts. We like to consider the data in an analytic framework to test whether the rumour is plausible and even viable. Call it a smell test, but we can try and use the data to answer some simple questions:

1. Does the underlying demand or network flows support the potential route?

2. Is it operationally possible from a fleet availability or aircraft performance perspective?

3. What is the airline’s history on the route and what have they been saying?

4. What is the opportunity cost or trade-off?

We can’t always answer all these questions as we often won’t have enough data, but using the framework is a very helpful smell test!

For example, a few months ago there was a rumour that Qantas are planning to start two Brazilian routes, the first during the 2nd half of 2026 with Rio being flown as an extension of the Sydney-Santiago flight on the B787-9, and during the 2nd half of 2027, a non-stop to Sao Paulo on the A350. We immediately discounted the rumour utilising this framework, for example:

1. In the 2024/25 year, the total Australia-Brazil market was 68,400 inbound passengers, amounting to an average of 187/day (according to ABS data). This is relatively small and compounded by Brazilian point-of-sale bias (ABS data tells us 62% are visitors and only 38% Australian residents). Furthermore, Brazil is covered through Qantas’s codeshare partnership with LATAM with competition from other stopover carriers limited by geography.

2. Fleet time on the B787-9 is scarce and even precious, and whenever B787-9 time has become available in recent years it’s quickly been applied to Europe (SYD-PER-FCO/CDG) and North America (e.g. SYD-YVR, BNE-LAX and MEL-DFW). Furthermore, the additional block time as an onward leg from Santiago to Rio would require a reorganisation of the schedule adding 11 hours block time.

3. The 1st A350 is arriving in late 2026 with deployment on non-stop Sydney-London from the start of the NS27 season (end of March 2027). Non-stop Sydney-London will require at least 3 aircraft, and with the 4th A350 expected by end of FY27, it seems unlikely that something like Sydney-Sao Paulo would be flown in 2027 and prioritised over non-stop Sydney-London, Melbourne-London and Sydney/Melbourne-New York. The latter routes are what Qantas have been telling us they’ll be flying.

Enough ranting, what about Jetstar and Sri Lanka?

So back to the rumour: Jetstar will start Melbourne-Colombo with 3x/week flights on the B787-8 on 28 August 2026.

It appeared on social media a few weeks ago, published by a Sri Lankan source. And it’s very specific which can be a sign of credibility or equally of a clever charlatan, so lets take a look and see if it ticks the boxes:

1. At a group level, Qantas currently access Colombo through a codeshare on SriLankan Airlines and Cathay Pacific (QF code on UL and CX). Interestingly, they don’t codeshare on SriLankan’s non-stop flights from Melbourne and Sydney to Colombo, preferring to carry their passengers on their own flights to Singapore and Hong Hong and codesharing on SriLankan’s flights from Singapore and Cathay Pacific’s flights from Hong Kong.

2. Furthermore, between November 2024 and their closing, Jetstar Asia flew 5x/week Singapore-Colombo with a split schedule to provide efficient connections to/from Qantas and Jetstar’s Australian flights at Singapore. We previously argued that most of Jetstar Asia’s lost connecting capacity was covered by Jetstar’s increasing direct capacity into Asia (e.g. significant increases in flights to Bangkok and Phuket) or through partnerships (e.g. Malaysia Airlines codeshare). While the latter is also true for Sri Lanka given the codeshares with SriLankan and Cathay, it reinforces the group’s previous interest in servicing Colombo on their own metal. And Jetstar have teased a route to Sri Lanka when speaking about the B787 refurbishment.

3. The timeline also aligns with Jetstar’s B787 refurbishment plan. One of the significant limitations of Jetstar’s long haul ambitions is the lack of crew rest areas on the B787-8. With westbound Melbourne-Colombo running at about 10.5 hours, crew rest facilities will be a much needed boost. According to Qantas’s full year results presentation, Jetstar expect the first refurbished B787-8 with crew rest installed around April 2026. We’re not 100% sure if this is when work starts or will be completed, but it’s likely that even if this is the start, there will be more than one frame ready by August.

4. Some might question if Jetstar have fleet availability given that the B787 fleet is very well utilised at present. We can assess future fleet availability by comparing the schedules for September 2025 and 2026 (first full month after the proposed start date). Looking at ADS-B data, Jetstar operated 269 return B787 long haul sectors during September 2025, accounting for 269 of 330 available fleet days (Jetstar have 11x B787s and September 30 days). Furthermore, 25 days were occupied by heavy maintenance (excluding normal line and base maintenance). For September 2026, the currently loaded schedule shows 274 scheduled return B787 long haul sectors. This doesn’t suggest that there’s much scope, but there’s a catch! During September 2025, Jetstar operated 12x return Sydney-Honolulu sectors that have since been cancelled. While it’s difficult to predict heavy maintenance schedule, the Sydney-Honolulu fleet time matches the need for 3x/week Melbourne-Colombo, even if an aircraft is assumed to be in heavy maintenance.

5. The final consideration is the market? It’s all good and well arguing that the group would like to replace the lost Jetstar Asia capacity and that there’s fleet availability, but the economics of a longer non-stop routes dictate a substantial market need. Also, why Melbourne? Is there anything in this that suggests more credibility than if the rumour were flights from Brisbane, Perth or Sydney? This requires a more detailed examination and as usual, we brought the receipts …

Australia-Sri Lanka in numbers

The Australia-Sri Lanka market has grown consistently over the last decade and a half, from just 47,310 in 2009/10 to 170,520 in 2024/25, and increase of 260%. Notably, growth has been led by outbound travel from Australia to Sri Lanka, increasing by 90,500 per year (32,800 to 123,300 per year) compared to 32,710 by inbound travel from Sri Lanka to Australia (14,510 to 47,220 per year). This growing outbound demand is driven by a large Sri Lankan immigrant community in Australia (more about that later).

As previously mentioned, SriLankan Airlines already operates to Australia. At present, the market is served by daily Colombo-Melbourne and 4x/week Colombo-Sydney flights. They are a relatively new entrant to the market, having begun flying to Australia in 2017.

As they’re the only airline flying direct flights between Australia and Sri Lanka, they represent the entirety of the BITRE passenger flows on the route. As the figure below shows, prior to the pandemic passenger arrivals on direct flights between Australia and Sri Lanka were smaller than the total arrivals estimated by the ABS data. This indicates that a net transit of passengers via third countries (e.g. Singapore and Hong Kong). However, in the immediate aftermath of the pandemic, this trend reversed as SriLankan Airlines were (anecdotally reported to have been) carrying significant net transit traffic to India. This trend how now normalised. Further, demand for travel between Australia and Sri Lanka shows far greater seasonality than direct passenger flows.

This is visualised more easily by showing just the monthly net transit figures (i.e. the difference between ABS and BITRE on a monthly basis). While the market has normalised, there’s a significant reliance on 3rd country transit for travel between Australia and Sri Lanka. It’s important to remember that these data represent net transit, so offsetting onward connections (e.g. onwards from Colombo to India, or onwards from Melbourne to Auckland). For example, in January 2025, there were 10,123 net transit passengers of 23,770 total arrivals. Meanwhile, a 3x/week Jetstar flight would provide a little more than 4,000 seats each month, highlighting that additional direct capacity can be absorbed. So the market is probably substantial enough, although that doesn’t imply anything about yields and the financial viabilty.

And Melbourne?

This is the point that closes the deal. As previously mentioned, there is a considerable Sri Lankan immigrant community in Australia. ABS data shows 172,800 Sri Lankan born residents in Australia in 2024, making it the 10th largest foreign born population and accounting for 2% of the foreign born population of Australia.

Furthermore, the Sri Lankan community in Australia is well-established and holds substantial ties to the old country. 63% of the Sri Lankan born population has been in Australia since before 2011, so much more established than more recent migration waves from China, India or the Philippines. This highlights the high proportion of VFR traffic that would be more consistent over time than purely outbound Australian tourism. So it’s not just about young Aussies going to find themselves while surfing in Arugam Bay.

The 2021 census data helps us understand this further, indicating that 51% of the Sri Lankan born population in Australia resides in Victoria (for our foreign readers, Melbourne is the capital of Victoria). By comparison, New South Wales (Sydney) and Queensland (Brisbane) account for 24% and 9%, respectively, highlighting why SriLankan airlines favour Melbourne over Sydney, and reinforcing the credibility of the rumour.

So yes, the rumour is plausible

We probably wouldn’t bet on it happening, but the rumour seems plausible. It’s possible that Jetstar have had discussions with local authorities which included substantive discussions about timelines and schedules. This could easily be misinterpreted as a commitment (call it broken telephone). However, such details are typically only discussed when planning is at an advanced stage.

But why hasn’t it been announced? Most carriers have already loaded schedules for August and September 2026, and even Jetstar have loaded their August and September schedules. So if it’s credible, why haven’t they just announced the route yet?

New routes to countries a carrier doesn’t operate to can be more complex than just adding capacity. In addition to negotiating with multiple local service providers, carriers are increasingly seeking (financial) assistance from local authorities and route development agencies to support start-up costs and/or provide incentives. Jetstar are no different and they may just be seeking to maximise these prior to announcing. Once you announce, your leverage is lost!

Update: the data used in this analysis is available on our Capacity Tracker dashboard. It’s easy to use and pretty fun to play with, allowing you to filter data by any combination of airline, Australian destination city, or international city and country of origin.

Last week, we published a poignant Chart of the Week showing how international airline capacity to Australia has finally returned to its pre-pandemic baseline. In the 12 months to July 2025, international reached 24.8 million seats, slightly exceeding the 24.7 million seats in the 12 months to January 2020. It took five and a half years to recover what was destroyed in a few weeks by the COVID-19 pandemic.

Given the length of time its reasonable to expect that the distribution of that capacity would’ve changed significantly, augmented by the dramatic structural shifts in the global economy. We’ve observed significant changes in the airline landscape with some new carriers having appeared or grown their Australian presence, and several notable casualties. Travel patterns have also changed, with the popularity of several destination increasing while other historically important destinations have suffered from slow recoveries and macroeconomic shocks.

As promised last week, let’s delve into the trends, first by looking at the distribution of that capacity on the Australian side of things and then the countervailing countries before looking at the evolution by airline. The BITRE data we use tracks capacity utilising inbound seats. We’ve included a methodological note at the end of the article to explain how we handle the data since there are some quirks as to how capacity is measured for multi-stop flights.

Australia

Perth’s performance is just stellar! There’s no other way to describe it. The net increase of 595,551 is actually a slight over estimate though as Qantas’s Sydney-Perth-Rome and Paris flights can’t easily be split between Sydney and Perth, and as the methodological note highlights, we make the simplification of adding it all to Perth. However, the maximum over estimate is 45,953 seats, leaving Perth with at least an increase of 549,598 seats. So where does Perth’s gain come from?

Almost everywhere, but with a strong emphasis on SE Asia including sizeable increases from Indonesia (AirAsia, Batik and TransNusa) and Singapore (Scoot), new Jetstar flights to Singapore and Bangkok, Qantas’s new Paris and Rome flights, and new carriers from the Philippines (Philippine Airlines) and Vietnam (VietJet and Vietnam Airlines).

Melbourne and Canberra also contributed positively, with Sydney and Adelaide showing declines. The latter may be considered temporary with the late returns of Cathay Pacific and Emirates not being fully accounted for in the data at this point.

The net decline of 361,986 seats to the Brisbane-Gold Coast-Sunshine Coast area is concerning. Gold Coast’s decline was led by the loss of Jetstar’s Tokyo and Seoul flights to Brisbane (102,845 seats) and the loss of other long haul services including Air Asia’s Kuala Lumpur flights (101,790 seats) and Scoot’s Singapore flights (73,238). Given this it’s maybe surprising that the number isn’t larger!

Brisbane posted a net decline of 149,761 seats, despite gaining 170,850 seats from taking over Jetstar’s Tokyo and Seoul flights. This highlights how much worse Brisbane would’ve fared had it not been for attracting those flights from Gold Coast and several other high profile gains include additional Jetstar flights to Bangkok and Osaka, and the arrival American, Delta and United. So where did it go wrong for Brisbane?

Virgin Australia! Simply put, Brisbane has borne the brunt of Virgin’s long haul decline, accounting for a net loss of 404,928 seats. Brisbane would’ve observed a significant increase had it not been for Virgin’s demise. That said, excluding Brisbane, Virgin’s lost 699,690 seats with Sydney leading the way with 479,209 seats. Without Virgin’s demise, Sydney would’ve also observed a net increase of 294,900 seats. Virgin cost Melbourne 215,840 seats that would’ve resulted in a large net gain of 312,854 seats.

Rest of the world

We’ve seen exceptionally large swings in the distribution of capacity by country. Few of these should surprise readers since we’ve discussed many individual country trends over the last two years.

For example, it shouldn’t surprise anyone seeing New Zealand generating the largest decline as we covered in a deep dive last month that focused on the impact of Virgin’s significant withdrawal of capacity from Trans Tasman flights combined with Air NZ’s fleet challenges.

Seeing the UAE might surprised some, but this predominantly the result of Etihad’s significant withdrawal of capacity rather than Emirates’s slow return of some capacity (e.g. only having recently returned to Adelaide). Hong Kong has suffered as a result of weak demand and Cathay Pacific’s very slow recovery, while China has also suffered from a very slow recovery.

Unsurprising, the United States shows significant capacity declines although the decline is less consequential in relative terms given higher baseline capacity. We’ve also discussed the weakness of the Australia-US market, hurt by the strong US dollar in a market dominated by Australian point-of-sale.

Seeing Malaysia in the group is somewhat surprising but likely overstated by Malaysia Airlines’s temporary operational challenges in late 2024 and early 2025, combined with AirAsia’s erratic capacity. We expect this to recover much quicker than other declines. Pacific Islands have also seen declines in capacity. While these are generally small individually and in absolute terms, it’s very significant in aggregate and relative terms.

On the positive side, Asia is booming! That’s as long as we exclude China, Hong Kong and Taiwan. Indonesia continues to thrive, driven by outbound tourism to Bali and growing inbound tourism, business, educational and VFR travel. Capacity is limited by BASA constraints, however growing demand has led Australian carriers to utilise creative ways to increase capacity and several Indonesian carriers are stepping up.

Vietnam is also booming, enabled by ambitious Vietnamese carriers and an expanded BASA. Japan is the probably the hottest tourist destination for Aussies, while Singapore and Thailand are old stalwarts with strong bidirectional travel and the latter increasing its already established reputation as a critical connecting hub. India’s growth has also been phenomenal with the absolute increase masking larger relative growth in the market, aided by Qantas’s return to the market and Air India’s ambitions.

Other notable gains include Qatar, the Philippines and Turkey. This is driven by Qatar Airways’s ambitious growth as a connecting point to Europe. Expanded flying by Philippine Airlines, Cebu Pacific and Qantas is supporting growing bi-directional travel with large Filipino immigrant communities in Australia and growing Australian business investments in the Philippines. And Turkey on the back of Turkish Airlines’s recent entry into Australia.

And by airline

This is probably the most interesting section for many readers. Not to anyone’s surprise, Virgin leads the way with a near catastrophic loss of capacity. While this will moderate in the coming months as Qatar’s Virgin’s Doha capacity is fully implemented and reflected on an annualised basis it’s still expected to generate a long term capacity decline.

It’s hard to conceptualise the impact of Virgin’s dramatic decline in international flying as few airlines had the ability to replace that capacity. And it’s not just long haul flying. As previously highlighted, it’s had a particularly large impact to New Zealand and the US where other airlines were unable to fill the gap quickly. In the case of the US, United Airlines certainly tried along with the American and Delta, but despite their large capacity increases, US aggregate capacity still declined, reinforcing the scale of Virgin’s decline.

Etihad’s close call with death shows itself here, although this was foreshadowed earlier by the UAE’s capacity decline and as is now obvious, Emirates’s slowish return is relatively minor. Chinese, Hong Kong and Taiwanese carriers feature heavily, as do Malaysia Airlines and Air Asia, as expected by the decline in capacity from Malaysia noted previously.

Two very interesting juxtapositions are relevent: Qantas and Jetstar, and Singapore Airlines and Scoot. On face value, one might critique Qantas and Singapore for their capacity declines (126,612 and 69,268 seats, respectively), however when evaluated at a group level, net increases of 859,959 and 122,356 seats, respectively, were observed (when including Jetstar Asia, Jetstar’s net increase was smaller at 815,319 seats). This highlights a significant capacity pivot from mainline full service carriers (FSCs) to low cost carriers (LCCs) at both groups.

Interpreting this as increasing demand or preference for LCCs and declining demand for FSCs is overly simplistic and incorrect. Rather, it represents a deliberate strategy by both Qantas and Singapore to segment lower yielding traffic onto a lower cost base and lower value product, while concentrating higher yielding traffic on mainline brands, supported by declining cabin density on mainline aircraft. In layman’s terms, we’re seeing mainline aircraft with fewer economy class seats, replaced by more and improved premium economy and business class seats (and sometimes first class). In effect, it allows mainline to have a stronger focus on increasing yield without having to balance the product and cost base for lower yielding traffic. This strategy is nuanced and is both under appreciated and misunderstood.

Other growth leaders are somewhat unsurprising given the previous country-level discussion. Standout performers include new entrant VietJet leading the way and Batik driving growth from Indonesia, and Vietnam Airlines, Air India, and ANA supporting growth from Vietnam, India and Japan, respectively. Interestingly, several Chinese carriers showed strong growth, offset by larger declines by others, generating a bit of movement in the deckchairs.

Conclusion

Clearly, the story is Virgin with a touch of Perth and Asia. Oddly, they have very little overlap with Perth benefiting from their very limited exposure to Virgin’s international capacity, while Virgin hurt Brisbane badly, while also undermining Sydney and Melbourne’s recoveries. This probably takes a bit of the lustre off Perth’s win!

Asia has been a clear bright stop with significant growth in traditional and emerging markets, with China, Hong Kong and Taiwan being exceptions, but this shouldn’t detract from the exceptional growth elsewhere in the region, particularly Indonesia, Japan and Vietnam. This all while the US continues to suffer at the hands of the strong US Dollar and Virgin’s demise.

Methodological annex

These data are drawn from BITRE and follows inbound capacity only. We use inbound as a proxy of the total to reduce the data analysis burden. Unlike passenger flows, what comes in must go out when measuring capacity.

A challenge when measuring capacity are flights with multiple stops, this includes co-terminalisation (e.g. Qantas’s Sydney-Perth-Rome) and 5th freedom flights (e.g. BA’s London-Singapore-Sydney). In the case of the former, we apply the capacity all to the most significant point (e.g. Qantas’s Sydney-Perth-Rome is all applied to Perth and Rome). In the case of the 5th freedom flights we apply it to the final destination (e.g. BA’s London-Singapore-Sydney is all applied to London and Sydney).

When 5th freedom flights fly through Australia (e.g. Emirates’s Dubai-Syndey-Christchurch), we apply the capacity fully on both legs, meaning the capacity is fully available on Dubai-Sydney and Sydney-Christchurch as if they were seperate flights.

This morning, BITRE released July 2025’s international aviation data, revealing an important milestone: international airline seat capacity has finally reached its pre-COVID level! In the year to July 2025, airlines flying international routes to Australia provided 24.8 million seats compared to 24.7 million seats in the year to January 2020.

While capacity recovered quickly in the immediate aftermath of the pandemic, the recovered slowed significantly in the last year. By the end of 2022, inbound capacity had reached 48% of the January 2020 baseline, and 84% by the end of 2023 and 95% by the end of 2024. It’s been like watching paint dry during 2025!

A bigger and more intriguing question is how the distribution of this capacity has evolved in the post pandemic period. This includes the distribution of this capacity within Australia (between different airports), by airline, and by city and country of origin. There have been dramatic shifts in the distribution of this capacity over the last five years, and we’ll share more on that in the coming days …

In case you missed it, our last chart of the week looked at how Gulf carriers came to dominate the Australian-UK market over the last two decades. Gulf carriers now dominate the market, accounting for nearly half of the market. Qantas and BA have been the major casualties, although Qantas have done well to recover and increase their market shares in recent years. Transit carriers like Malaysia Airlines and Singapore have suffered large losses.

The more recent emergence of Qatar has thrown a spanner in the works as they now compete effectively with other Gulf carriers, having won significant chunks of market share from Emirates and Etihad. No carriers have been safe!

Over the last year, we’ve published 43 articles. We had aimed for one a week but this wasn’t possible given the extensive time taken to produce long-form content and life getting in the way. We welcomed Analytic Flying Jnr in January and suffered the sudden loss of a parent a few months later. At times life was a distraction, and at other times Analytic Flying was a great distraction for us. Of the 43 articles, 15 were shorter “chart of the week” pieces with interesting data and less analysis.

We also promised a more global focus and we think we’ve delivered. Several articles over the last year focused entirely on non-Australian topics while others were global topics with relevance to contemporary issues in Australian aviation. In the former category, we spent some time looking at Air New Zealand’s fleet shortages and comparing the strategies behind the global connecting hubs. We delved into the Indian market, including extensive work on the A321XLR at IndiGo. One reason we’ve focused on India is its phenomenal growth, dynamism and potential!

In the later category, two recent analyses come to the fore. We used US data to compare the operating costs of regional and narrowbody jets, highlight how regional jets are misunderstood by the public, and how they wouldn’t have saved Bonza. We looked at the EU’s compensation scheme, showing how it generates obtuse incentives for airlines. We hope that it’ll help inform debate on the Aviation Consumer Ombuds Scheme.

This follows on from other novel analyses that included a comprehensive review of Emirates’s hub strategy, utilising their Australian network as an instructive example to explain how and why they’re willing (required?) to schedule flights with 18 hour turnarounds in Australia.

That said, our two most enjoyable analyses this year are two that also went relatively unnoticed, or at least didn’t garner nearly as much attention relative to the effort that went into them. This just reinforces the joy that we have gained working on Analytic Flying over the last two years!

The first of these analyses was a detailed tracking of Qantas’s fleet return after the COVID-19 pandemic. While the results may not have been earth shattering, it highlighted how aircraft availability was the real determinant of capacity. This database led to several follow-on analyses.

The second of these may have been our Jerry Maguire moment, albeit a subtle one. It focused on airline profitability and why being profitable isn’t good enough in a sector with unusually high capital costs. Hopefully it helped contextualise why the sector is relatively concentrated and why many new entrants will struggle and likely fail. They simply don’t have enough capital or alternatively poor margins and returns will undermine their ability to afford that capital when markets becomes more competitive.

What next?

We keep reminding readers that we do this because we enjoy it, and we have to say that we’ve really enjoyed the last year! Analytic Flying isn’t a business. There are no ads or credit card sign-up links on our website. We don’t begrudge websites that do, but this isn’t our vibe. Our funding model is simple: everything is free and we encourage readers to sign-up for a subscription to help us track impact.

Subscription is free, but we also have a voluntary paid subscription option for readers who would like to contribute to our running costs. Thank you to those who have signed-up for a paid subscription. We appreciate and value your support! Our model has constraints as it doesn’t afford us a professional staff to churn out high frequency content like short form news, proof read articles, or purchase access to industry databases. But we don’t think that that’s what you’re coming for!

Something that’s irritated us over the last few months is the increasing use of our content by other platforms, often without attribution. We’re don’t mean a linked reference to Analytic Flying (that’s really good actually), rather paraphrasing our work in other “original” content without attribution. Moreover, doing this with the explicit purpose of developing monetising content.

We’ve spoken to a few new media experts for advice. We don’t want to put Analytic Flying behind a paywall as this defeats the purpose of the project. Instead, we’re proposing an “ethical paywall”. It’s pretty simple: if you use Analytic Flying for work or any professional or commercial purpose then sign-up for a paid subscription. If it’s just personal use then carry on as you are, whether that be a free or paid subscription. An ethical paywall will hopefully ensure that important information remains accessible to the public for personal use.

While Analytic Flying isn’t a business, more paid subscriptions will allow us to do more. That includes more frequent and regular content, and utilising industry data in addition to publicly available data (these databases are very expensive). Many readers have also been encouraging use to develop alternative content streams like video content to podcasts, but these are well outside our capabilities at present.

Over the last few months we’ve seen Qantas and Jetstar come out swinging with new routes and capacity increases across its trans-Tasman network. This includes Qantas’s forthcoming returns to Perth-Auckland and Adelaide-Auckland in addition to seasonal increases across its existing network. In addition to increases across its existing network, Jetstar has also added several new routes including Cairns-Christchurch, Gold Coast-Hamilton, Gold Coast-Dunedin, Sydney-Hamilton, and Brisbane-Queenstown. In 2025/26, Qantas and Jetstar will add more than 800k seats to its trans-Tasman schedule, a 20% increase compared to 2024/25, already their highest ever annual capacity. While we always encourage capacity increases, the reasons for these increases are concerning, painting a dismal picture for competition on trans-Tasman flights in the coming years. Let’s explore …

Is demand increasing?

In 2024/25, the ABS reported 2.8 million arrivals in Australia from NZ, an increase of 5% over the previous year, but still 2% below the pre-pandemic peak in 2018/19. Meanwhile, BITRE reported 3.4 million arrivals in Australia on flights from NZ in 2024/25, a 5% increase over the previous year, yet still 6% below the pre-pandemic peak (also in 2018/19). Notably, the pre-pandemic period showed consistent growth of 32% (ABS) and 33% (BITRE) between 2009/10 and 2018/19.

Given the slow pace of recovery in trans-Tasman demand, Qantas and Jetstar’s capacity increases seem rather bullish, provoking several questions:

1. Is the slow pace of recovery due to a lack of supply, or is demand just weak?

2. Why are passenger flows reported by BITRE so much larger than ABS’s arrivals?

The questions are proposed in order of importance, however answering the second helps us with the first. The difference between ABS and BITRE data is attributable to transit traffic, i.e. passengers traveling to/from NZ through Australia (e.g. Auckland-Melbourne-Bali), and passengers traveling to/from Australia through NZ (e.g. Brisbane-Auckland-Los Angeles). In 2024/25, BITRE reported 571k more arrivals than ABS, meaning that there were approximately 571k from passengers traveling to/from NZ through Australia than there were passengers traveling to/from Australia through NZ. The pattern of this flow has been consistent over time, however the absolute and relative magnitudes have varied.

Net transit traffic has declined in both absolute and relative terms. In 2024/25, net transit accounted for 17% of the total traffic on trans-Tasman routes, having declined from 21% in 2018/19 and a peak of 23% in 2014/15 and 2015/16. Several factors may account for the decline, but most likely due to weaker demand between Australia and the US, and Air NZ’s fleet challenges that are limiting their ability to fulfil historic NZ-US capacity. We’ve covered these topics extensively (see below).

While this helps explain the decline in BITRE flows, it doesn’t explain the decline in ABS flows which are indicative of a decline in O&D traffic. Conceptually a reduction in Australia-US transit traffic via NZ opens capacity for O&D traffic, however the decline in the proportion of net transit traffic indicates that more capacity has been lost from the market than the decline in net transit traffic. This reinforces the importance of our first question, raising the prospect that demand been constrained by a lack of supply?

Tight market supply would lead to higher prices and reduce demand. This is difficult to directly measure, however increasing passenger load factors and prices support this hypothesis. Excluding 5th freedom carriers (see explanation below why we exclude them), the trans-Tasman aggregate load factor increase from 76% in 2018/19 to 83% in 2024/25.

Meanwhile, Air NZ reported unit revenues and yields increasing 41% and 30% between 2019/20 and 2024/25 on trans-Tasman and Pacific Island routes (Qantas and Jetstar don’t disaggregate data sufficiently to generate a comparable estimate). Since the Pacific Islands represents a fraction of their capacity compared to trans-Tasman, this provides strong evidence of both higher load factors and higher prices. We’ll come back to this later, but evidence of higher loads and prices combined with declining transit traffic supports the argument that there’s been a supply crunch!

Methodological comment: why segregate 5th freedom flights?

There are several 5th freedom carriers providing trans-Tasman capacity, e.g. Emirates’s Dubai-Sydney-Christchurch and LATAM’s Santiago-Auckland-Sydney (although the latter is ending soon). These flights carry traffic on all three sectors, e.g. Dubai-Sydney, Dubai-Christchurch and Sydney-Christchurch. But there’s a quirk in the BITRE data: it measures passenger flows on both Dubai-Sydney and Sydney-Christchurch legs separately, but it doesn’t measure passengers on the Dubai-Christchurch leg. This means we don’t know how many seats are actually available for sale for the 5th freedom Sydney-Christchurch leg and BITRE data will underestimate the load factor on 5th freedom legs.

Who is responsible for the supply crunch?

Qantas and Jetstar certainly aren’t responsible for the capacity crunch in the trans-Tasman market. They’ve added 472k additional seats in 2024/25 compared to 2018/19, an increase of 13% (6% and 32% individually). Meanwhile, Air NZ provided 211k (6%) fewer seats over the same time period, meaning a net increase at a market level, so who are we forgetting? We have short memories, but it’s easy to forget that prior to entering voluntary administration, Virgin Australia had a vast trans-Tasman network with 82 to 119 weekly flights each way connecting 13 city pairs.

They’ve since significantly downsized trans-Tasman flying, returning to only 3 routes to/from Queenstown, flying 27 to 36 weekly frequencies. Virgin’s capacity reduction has removed 1.3 million seats from the market (79% reduction). 5th freedom carriers have withdrawn 289k fewer seats from the market (28% reduction), resulting in a combined net reduction 1.4 million seats (14% reduction), reshaping the market!

As measured by passenger market share, Qantas and Jetstar now combine for 48% of the market, up from 37% in 2018/19. Despite declining capacity, Air NZ’s market share increased from 39% to 43% as their decline was slower than the aggregate market. Meanwhile, Virgin’s market share plummeted from 17% to 5%, reinforcing the impact of their capacity reductions.

5th freedom carriers also saw their market share decline from 7% to 5%, however this obscures the more consequentially decline from 17% in 2016/17. This larger decline is attributable to Emirates’s cancellation Brisbane-Auckland and Melbourne-Auckland in March 2018. A secondary consideration is whether we should consider Emirates’s capacity alongside Qantas and Jetstar, as they have a joint venture that extends to trans-Tasman routes. When combined, their market share has increased from 40% to 50% between 2018/19 and 2024/25, further highlighting their increasing market dominance in light of Virgin’s declining trans-Tasman capacity.

We’ve previously argued that the ACCC should’ve placed stronger conditions on the renewal of the Qantas-Emirates joint venture in 2023, with specific emphasis on trans-Tasman routes. Generally, joint venture approvals focus on ensuring capacity doesn’t decline on overlapping routes and/or networks as a result of the cooperation between the airlines, however the decline in combined market share from 43% to 40% between 2016/16 and 2018/19 (coinciding with the cancellation of Emirates’s Brisbane-Auckland and Melbourne-Auckland in March 2018) suggests that Qantas and Jetstar’s capacity increases during that time were insufficient to cover Emirates’s reductions.

Looking ahead

As noted at the outset, Qantas and Jetstar will add 800k more trans-Tasman seats in 2025/26. All else held constant, we expect their share of capacity will increase from 48% to 52%, or 52% to 56% including Emirates. In 2018/19, this was just 36% and 40%, reinforcing their increasing dominance. Furthermore, we should be careful not to ascribe blame as they aren’t responsible for Virgin and Air NZ’s challenges, and the counterfactual of them not adding this capacity would be an even larger supply crunch leading to even higher prices!

Things Can Only Get Better?!?!

Listening to the 1980s pop anthem “Things Can Only Get Better” gives us a sense of optimism about overcoming challenges. Unfortunately, listening to catchy songs doesn’t change the fact that things aren’t going to get better, or at least not anytime soon! Our thesis has three key points:

1. Air NZ aren’t in a position to counter Qantas and Jetstar’s capacity increases.

2. Sadly, 5th freedom carriers aren’t coming to save us.

3. Virgin don’t have any incentives to return trans-Tasman capacity in light of the nuances regarding their codeshare on Air NZ’s trans-Tasman routes.

Firstly, Air NZ’s fleet constraints aren’t going to be solved anytime soon. During their results presentation last month they confirmed the “Grounded aircraft position (is) expected to improve slowly, with majority returned to service by end of calendar 2027.” The operative word is “majority”, meaning that fleet constraints are likely to affect them through 2028. Even then, Air NZ are still short of their own historic benchmarks, nevermind filling the large gap that Virgin has left.

Secondly, 5th freedom carriers aren’t going to suddenly increase trans-Tasman capacity. Instead they’ll continue to reduce trans-Tasman flying, despite higher prices and yields. The most recent casualty is LATAM who will exit trans-Tasman 5th freedom routes in late 2025, favouring more non-stop capacity to Australia and NZ. This is a long term trend that we analysed in detailed a few years ago.

Finally, Virgin aren’t likely to return significant trans-Tasman capacity in the near future. Amidst aircraft delivery delays and historically poor performances on trans-Tasman routes, they’ve sought a codeshare on Air NZ instead (doesn’t extend to Air NZ’s domestic services, nor to Air NZ codesharing on Virgin’s trans-Tasman or domestic flights). This was approved by the ACCC in June 2024 and gave them access to to Auckland, Christchurch and Wellington, thereby helping them compete more effectively with Qantas’s network.

The approval included a nuanced technical point: Virgin only sought approval for codeshares on non-overlapping routes, thus excluding codeshares on Air NZ flights on routes to/from Queenstown. However, this is also forward looking meaning that if Virgin was to enter a new trans-Tasman route already served by Air NZ (e.g. Sydney-Auckland) they would cease codesharing on that overlapping route. On face value, this is very reasonable as codesharing on overlapping routes would limit competition on that route. However, it’s overly simplistic, ignoring the incentives for a new entrant on a high frequency route. Let’s consider the prospect of Virgin entering Sydney-Auckland:

At present, Qantas and Air NZ each have 4-5x/day flights (excluding QF3/4 that continues onwards to New York). Their schedules are distributed throughout the day, offering passengers a wide variety of times and connecting options, enabling them to generate higher yields than they would from similar capacity but concentrated capacity.

Without several daily frequencies Virgin would struggle to attract higher yielding passengers, resulting in lower yields than Qantas and Air NZ but without any cost advantage. To compete effectively with Qantas and Air NZ, Virgin would need to provide both morning and evening departures from Sydney, requiring at least 2x/day flights, and 3x/day if they aim to provide morning and evening departures from Auckland. That’s a lot of capacity for a new entrant to commit where they’d typically prefer to enter a new route with less capacity, limiting risk and building capacity over time as the route develops.

SYD-AKL
QF 141  7:05 12:12 D  B738
NZ 110  7:45 13:00 D  A32N
QF 143  8:55 14:05 D  B738
NZ 102  9:55 15:10 D  A32N
NZ 104 11:45 16:50 D  B77W
QF 145 13:05 18:15 D  B738
NZ 106 13:55 19:00 6x B789/B77W
QF 147 17:50 23:00 D  B738
QF 149 19:15 00:25 D  B738
NZ 112 19:15 00:30 D  A32N

AKL-SYD
QF 140  6:00  7:35 D  B738
NZ 101  7:00  8:45 D  A32N
QF 142  7:35  9:10 D  B738
NZ 103  9:00 10:30 D  B77W
NZ 105 11:10 12:40 6x B789/B77W
QF 144 13:15 14:50 D  B738
QF 146 16:05 17:40 D  B738
NZ 111 16:20 18:05 D  A32N
QF 148 19:15 20:50 D  B738
NZ 109 20:00 21:45 D  A32N

But it’s all or nothing for Virgin. If they were to enter the route 1x/day they would loose access to all of Air NZ’s 4-5x/day, while codesharing on overlapping routes would give them access to the alternative timings on Air NZ’s flights and help them maintain a competitive schedule with only 1x/day flights. Thus, the inability to codeshare on overlapping routes reduced Virgin’s incentive to enter any of these routes.

However, the ACCC approved Virgin’s application noting “concerns that the proposed conduct may reduce Virgin’s incentive to enter new trans-Tasman routes”. They indicated they were satisfied that it wouldn’t materially reduce Virgin’s incentive to enter new trans-Tasman routes after considering confidential information provided by Virgin on the revenue and margins they expected to earn from the codeshare compared to its current operations. That’s absurd!

Essentially, Virgin argued that their ability to add trans-Tasman capacity in the short run is constrained by overall capacity and that they’d earn higher revenue and margins elsewhere in their network. Airlines are far better placed than ourselves or regulators to decide where it’s best (most profitable) to allocate their scarce capacity. So we don’t contest this, however it’s not relevant to whether or not codesharing on overlapping routes would materially affect Virgin’s incentive to enter new trans-Tasman routes.

We might even go as far as using that position to argue lower revenue and margins on trans-Tasman routes compared to Virgin’s wider network means that codesharing on overlapping routes is neccessary for Virgin to even consider entering a new route like Sydney-Auckland. And it isn’t just Sydney-Auckland. Similar frequency dominated schedules are found on other routes including Melbourne-Auckland (Qantas 4x/day, Air NZ 4x/day), Brisbane-Auckland (Qantas 2x/day, Air NZ 4x/day), Sydney-Christchurch (Qantas 2x/day, Air NZ 2x/day), and Melbourne-Christchurch (Qantas 2x/day, Air NZ 2x/day).

We’ve going to leave you in limbo

This has been an unusually pessimistic analysis! Trans-Tasman routes have experienced a capacity crunch precipitated by Virgin’s significant withdrawal of capacity, leading to higher prices that have further surpassed demand. While systematic data on prices isn’t available, Air NZ’s yields have surged 30% since 2019/20, going a long way to explain Qantas and Jetstar’s significant capacity increases, while opportunistically exploiting Air NZ’s challenges to gain/increase their dominance. They’ll hold at least 56% of capacity in the coming year (compared to just 40% before the pandemic), leaving Air NZ with a lot of FOMO. If only they had the capacity to push into the market right now!!

We also shouldn’t be applauding Qantas and Jetstar. They’re not altruists, rather they’re following the money. But it begs the question why Qantas and Jetstar are choosing to allocate scarce capacity to trans-Tasman routes while Virgin are telling the ACCC that the market doesn’t look good?! We acknowledge that it’s easy to criticise from afar and this isn’t a shot at the ACCC. They don’t bare any responsibility for Virgin withdrawing so much capacity and the didn’t apply the non-overlapping condition. Virgin’s applied for it and they didn’t find any reason to exclude it. We must also consider the counterfactual that even if Virgin were allowed to codeshare on overlapping routes that they might not reenter these routes. Maybe we are just more worried about the our different reading of the strategy of the precedent it sets. The ACCC also couldn’t have contemplated Virgin’s strategy when they approved the renewal of the Qantas-Emirates joint venture. While we argued that the ACCC should have placed stronger conditions on the approval or limited its scope, it’s unlikely that this decision has affected Virgin’s strategy.

But where does this leave us? Are we just at the mercy of the market? Air NZ have strong incentives to push more capacity into trans-Tasman routes, however that won’t be anytime soon. In the interim, Qantas and Jetstar are filling the void but at the considerable long term cost of them increasing their dominant position that will give them greater pricing power going forward!

While somewhat outside the scope of this analysis, we do hope the ACCC will continue to track the trans-Tasman market, including following-up with Virgin to review how their estimates of revenue and margins play out. Would the ACCC reconsider the approval if market conditions are shown to be different? Even then, what can they do? As our counterfactual states, allowing codesharing on overlapping routes doesn’t mean they’ll actually do it. However, if market conditions are shown to be different would the ACCC go as far as revoking the approval to encourage Virgin to enter the market? Also, would they revisit the scope of Qantas and Emirates’s joint venture to encourage Emirates to reenter 5th freedom sectors, or reimpose it as a condition for future renewal?

Last month, Qantas loaded several charter flights from Brisbane and Sydney to Los Vegas in February and March 2026 in support of Rugby League Las Vegas. This annual event is entering its third year in 2026 and this will be the third time that Qantas have operated charter flights for the event. As reported by Aeroroutes, Qantas will operate four return flights, one from Brisbane and three from Sydney.

This is a significant increase compared to previous years, with only two flights in 2024 and 2025 (one each from Brisbane and Sydney). What’s immediately obvious from the schedule is that the return flights (1 and 2 March 2026) are a week after the outbound flights (24 and 25 February 2026). Qantas aren’t just going to leave four B787-9s sitting around Vegas waiting for the return, so how are they going to schedule this?

Outbound:

QF341 SYD LAS 13:00 07:40 B789 24FEB26
QF343 SYD LAS 15:20 10:00 B789 24FEB26
QF341 SYD LAS 15:20 10:00 B789 25FEB26
QF345 BNE LAS 20:30 15:50 B789 25FEB26

Return:

QF344 LAS SYD 20:00 06:20+2 B789 01MAR26
QF342 LAS SYD 23:25 09:45+2 B789 01MAR26
QF346 LAS BNE 19:40 05:00+2 B789 02MAR26 
QF342 LAS SYD 23:25 09:45+2 B789 02MAR26

Do they run empty legs, flying the the four B787-9s back to Australia empty after delivering their passengers on 24 and 25 February, and them position empty back to Las Vegas from Australia for 1 and 2 March? That’s obviously absurd, so what’s the plan?

Where the scheduling fun begins …

After a detailed look at Qantas’s forward schedule we’ve been able to work out how they’ll operate the positioning sectors to/from existing Qantas US destinations, while proactively removing the countervailing flights from the schedules 8 months ahead of time. It’s a stretch to consider these cancellations as opposed to just not scheduling these sectors given the advanced nature of the changes.

For background, Qantas operate multiple US routes, including daily flights between Brisbane, Melbourne, Sydney and Los Angeles, 3x/week flights between Sydney and San Fransisco, daily and 4x weekly flights between Melbourne, Sydney and Dallas, 5x/week flights from Sydney to New York via Auckland, and flights between Melbourne, Sydney and Honolulu.

The schedules of the Los Angeles and San Fransisco flights are shown below. We’ve excluded the Dallas, Honolulu and New York flights as these aren’t utilised for the positioning flights.

Outbound from Australia:
QF 15 BNE LAX B789 Daily
QF 93 MEL LAX A380 2x/week (M Sa)
QF 93 MEL LAX B789 5x/week (Su Tu W Th F)
QF 11 SYD LAX A380 Daily
QF 73 SYD SFO B789 3x/week (Tu F Su)

Return from the US:
QF 16 LAX BNE B789 Daily
QF 94 LAX MEL A380 2x/week (Th Sa)
QF 94 LAX MEL B789 5x/week (Su M Tu W F)
QF 12 LAX SYD A380 Daily
QF 74 SFO SYD B789 3x/week (Su Tu F)

A further challenge for Qantas’s scheduling team is that Sydney-Los Angeles flights and some Melbourne-Los Angeles flights are flown by the A380 rather than the B787-9, taking them out of consideration as the Vegas charters will be flown on the B787-9.

So here’s what’ll happen. On the outbound from Australia to the US, all the charters arriving in Las Vegas will position empty to Los Angeles the same day to fly originally scheduled Los Angeles departures to Brisbane and Melbourne, with the countervailing flights from Brisbane and Melbourne to Los Angeles cancelled and replaced by the charters to Las Vegas. Notably, the arrival times in Las Vegas and the late night departure times of Los Angeles to Australia flights mean that positioning flights will occur on the day and no time lost in the schedules.

• QF341 SYD-LAS (24 Feb) replaces QF93 MEL-LAX (cancelled on 24 Feb). Aircraft will then position LAS-LAX and then operate QF94 LAX-MEL (24 Feb).

• QF343 SYD-LAS (24 Feb) replaces QF15 BNE-LAX (cancelled on 24 Feb). Aircraft will position LAS-LAX and then operate QF16 LAX-BNE (24 Feb).

• QF341 SYD-LAS (25 Feb) replaces QF93 MEL-LAX (cancelled on 25 Feb). Aircraft will position LAS-LAX and then operate QF94 LAX-MEL (25 Feb).

• QF345 BNE-LAS (25 Feb) replaces QF15 BNE-LAX (cancelled on 25 Feb). Aircraft will then position to LAS-LAX and then operate QF16 LAX-BNE (25 Feb).

On the return flights from the US to Australia a week later, the charters from Las Vegas will replace cancelled flights from Los Angeles to Brisbane and Melbourne, as well as a San Fransisco to Sydney flight. The aircraft will position empty from Los Angeles and San Fransisco after completing from Brisbane and Melbourne in the case of Los Angeles, and Sydney in the case of San Fransisco.

Notably, positioning flights on the return leg are more complicated as late afternoon arrivals into San Fransisco (from Sydney) and Los Angeles (from Brisbane) mean that one QF15 BNE-LAX on 1 Mar has been retimed to depart more than 2 hours earlier, while QF73 SYD-SFO on 1 Mar will overnight before/after positioning from San Fransisco to Las Vegas.

• QF344 LAS-SYD (1 Mar) operates instead of QF94 LAX-MEL (cancelled on 1 Mar). Aircraft will position LAX-LAS after arriving that morning as QF15 BNE-LAX.

• QF342 LAS-SYD (1 Mar) operates instead of QF74 SFO-SYD (cancelled on 1 Mar). Aircraft will position SFO-LAS after arriving that morning as QF73 SYD-SFO.

• QF346 LAS-BNE (2 Mar) operates instead of QF16 LAX-BNE (cancelled on 2 Mar). Aircraft will position LAX-LAS after arriving that morning as QF15 BNE-LAX.

• QF342 LAS-SYD (2 Mar) operates instead of QF94 LAX-MEL (cancelled on 2 Mar). Aircraft will position LAX-LAS after arriving that morning as QF93 MEL-LAX.

Now, imagine what a fun job this is to manage this jigsaw puzzle?

Note: we started working on this piece in December 2024 when there was a lot of discussion regarding the implementation of a flight cancellation and delay compensation scheme in Australia. We kinda forgot about it, but picked it up again given that the Aviation Industry Ombuds Scheme is due to be released soon.

EU261 is the European Union’s Air Passengers Rights Regulation that is often view as a model for air travel regulation in Australia. A defining and popular feature of EU261 is the compensation that passengers are due in the case of cancelled or delayed flights. Who wouldn’t want to be compensated €600 if your flights was cancelled or delayed?

While this sounds enticing, few consider the unintended consequences of regulations. Simply put, “there’s no such thing as a free lunch”, and fixed cost compensation schemes will likely result in airlines optimising their operations to minimise compensation rather than minimising cancellations and delays. Let’s delve a little deeper into understanding how airlines are likely to respond.

The mechanics of EU261

Since EU261 is often refereed to as a model for compensation schemes, let’s look at how E261 works, and doesn’t work. You may be due compensation of €250. €400 or €600 if your flight is delayed, with the amount depending on the length of flight and delay.

1. For flights <1,500 km, you are due €250 if you arrive at your final destination more than 2 hours late;

2. For flights >1,500 km <3,500 km, you are due €400 if you arrive at your final destination more than 3 hours late;

3. For flights >3,500 km, you are due €600 if you arrive at your final destination more than 4 hours late.

Most proponents of EU261 often forget the words “may be due compensation” since there are a multitude of exemptions where passengers aren’t due compensation due to extraordinary circumstances. These are typically factors outside the airline’s control, like weather or air traffic control problems, but also safety. These exemptions are critical since regulators don’t want to encourage airlines to take unnecessary safety risks to avoid compensation.

As with many things within EU bureaucracy, there is plenty of guidance on how to interpret these exemptions. For example, mechanical faults are included but only pertains to the flight on which the problem occurred. If the flight you are on encounters a mechanical fault them you aren’t due compensation, but a delay to the following flight on the same aircraft would be due compensation. Another example is a strike by airline staff which isn’t considered extraordinary circumstances, but a strike by an airline supplier is.

In addition to delays, passengers may also be due compensation if a flight is cancelled. Again there’s a catch, as you’re only due compensation if you’re notified of the cancellation less than two weeks prior to departure. This means that airlines have carte blanche to cancel flights two weeks in advance.

If notified of cancellation between one and two weeks prior to departure, no compensation is due if passenger is rerouted so that they can depart no more than two hours earlier than scheduled, and arrive no more than four hours later than scheduled. And if notified of cancellation less than one week prior to departure, no compensation is due if passenger is rerouted so that they can depart no more than one hour earlier than scheduled, and arrive no more than two hours later than scheduled.

Note that the threshold doesn’t refer to when the flight is cancelled, but when the passenger is notified. Recall that this distinction is important in light of Qantas’s ghost flight case where the ACCC argued that Qantas’s failure wasn’t just the cancellation of the flights but the delay in customer notification.

Any compensation scheme needs cutoffs and thresholds

A key feature of any compensation scheme like EU261 are the cutoffs or thresholds, for example the length of the delay determining if compensation is due or how much compensation is due, or how long in advance a flight is cancelled. Why are these cutoffs and thresholds required?

Firstly, it’s not reasonable to apply the same compensation to a short delay versus a long delay, or a shorter cheaper flight versus a longer more expensive flight. This should be obvious as a 1 hour delay just isn’t the same as a 12 hour delay, nor is a 1 hour delay to a 1 hour domestic flight and a 1 hour delay to a 16 hour ultra long haul flight.

However, the cutoffs are based on discrete thresholds and ultimately arbitrary. A flight that covers a distance of 1,499km that is delayed 3 hours gets €250, but a 1,501km flight that is delayed 3 hours gets €400. Similarly, notification of a flight cancellation 15 days out isn’t due compensation, while 13 days out is. Once again, this two week line in the sand is arbitrary, but it has to be drawn somewhere.

Gaming the system #1: exploiting 2 week threshold

Air travel with legacy airlines can be booked up to 11 months in advance. That’s the typical timeline when airline reservations systems (at least legacy GDS) make inventory available for sale to the public. Some airlines, particularly low cost airlines have significantly shorter timelines for inventory availability.

Advanced reservations has benefits for both airlines and customers. It allows airlines to increase cash flow but also gauge demand, allowing time to adjust capacity to meet market demand. It also gives customers greater flexibility and choice, and most importantly, cheaper fares.

Hypothetical, let’s consider what would happen if airlines had to pay compensation for any cancellation, irrespective of how far in advance the flight is cancelled. This would increase their costs of making inventory available in advance and/or reduce their benefit of making inventory available far in advance.

Ultimately, it increases the risk of making inventory available in advance and their likely response will be to shorten the advance period, reduce capacity made available in advance including the availability of cheaper fares (since the relative risk is higher due to fixed cost nature of the compensation), or dripping capacity into the market. Furthermore, they might disproportionately apply these to flights that may be at higher relative risk, particularly lower yielding flights (again, due to fixed cost nature of the compensation). Any combination of these would be a poor outcome for passengers.

So imposing a cutoff is necessary but this is only the first part of the problem. Once a cutoff is imposed the arbitrary nature of the threshold generates incentives for airlines that they can (and will) build optimisation models around. Here’s a typical example of what might happen and it’s actually a real-world example that we encountered a few years ago:

• We booked a flight on 31 December 2023 for travel on 24 July 2024 on Wizz Air. The flight was booked 7 months in advance. The itinerary was Santorini to Tel Aviv, departing 10:35am, arriving 12:25pm; costing €497.98 for two passengers.

• On 6 May 2024, Wizz Air notified us that our original flight has been cancelled. They offer us an alternative flight on 22 July 2024 (departing 6:45pm). This is already 4 months after we made the reservations, so obviously we’ve already made other other reservations including other flights and accomodation, making leaving two days earlier not at all viable.

• So we have to opt for the full refund and make our own alternative arrangements. But since 4 months have passed and just 2.5 months remaining before the flight, alternative options are likely to be more expensive.

• Our best option was to rebook on Aegean Airlines, departing on 24 July 2024, departing 2:55pm, arriving 6:50pm, and flighting via Athens. This trip takes us a little longer and costs us €596.80.

So we’re out of pocket €98.82 and taken more than double the flight time, but because we’ve been notified more than 2 weeks in advance we’re not due any compensation. In this specific case they actually notified us 10 weeks in advance (nice of them to do so), and highlights a key point, that thresholds are arbitrary. A line in the sand has to be drawn somewhere, whether that be 2 weeks, less than 2 weeks, or more than 2 weeks.

Wherever these thresholds are drawn and whatever the penalties (i.e. compensation), they just become the parameters around which airlines make optimising decisions. Increasing 2 weeks to 2 months will just push airlines to cancel flights earlier. This will benefit passengers on flights that are cancelled but will also increase the risk for airlines and result in other unintended consequences.

But it gets worse: we started with the lease egregious example! The next examples show how it can get worse for passengers as airlines can (and will) find methods to minimise compensation due from delays.

Gaming the system #2: distribute delays

While the goal of a compensation policy like EU261 is to reduce or eliminate cancellations or delays, the use of compensation as a tool will actually result in airlines optimising their operations to minimise compensation. Once again, it’s about discrete thresholds. Let’s take the EU261 thresholds and apply them to the Australian domestic market in a practical example:

• Airline X has a Melbourne-Sydney sector scheduled to depart Melbourne at 10:00am, with a scheduled arrival time in Sydney of 11:30am. It’s due to be operated by an aircraft flying into Melbourne from Brisbane with a scheduled arrive in Melbourne of 9:00am.

• The airline’s operations centre learns that the inbound from Brisbane has been delayed 3 hours due to a mechanical fault. While the Brisbane-Melbourne flight isn’t due any compensation as its delay is due to a mechanical fault, this protection doesn’t extend to the following sector. However, it results in the delay of the Melbourne-Sydney departure from 10:00am to 1:00pm, with a new estimated arrival at 2:30pm instead of 11:30am.

• The resulting 3 hour arrival delay would result in passengers being due A$450 (€250) in compensation. Airline X will want to avoid paying this compensation. The goal isn’t to avoid the delay, but rather avoid the A$450 in compensation!

• So the operations centre starts looking for an alternate aircraft. Can they cannibalise an aircraft from another flight? Importantly, they’re not looking for an aircraft that can depart Melbourne at 10:00am, but anytime up to 12:00pm, as this will mean arriving in Sydney by by 1:30pm. As long as they arrive in Sydney by 1:30pm they’ll avoid paying compensation which is only becomes due if the flight is delayed by more than 2 hours.

• The operations centre find a Melbourne-Adelaide that’s scheduled to depart Melbourne at 12:00pm, arriving in Adelaide at 12:40pm. The airline takes this aircraft and transfers it to the Melbourne-Sydney flight. The flight departs Melbourne at 12:00pm and arrives in Sydney at 1:30pm. It’s arrived 2 hours late instead of 3 hours late. This is a better outcome for the Sydney bound passengers, but more importantly, Airline X doesn’t have to pay compensation.

• But what about the Melbourne-Adelaide flight? It now takes the aircraft originally scheduled for Melbourne-Sydney aircraft that arrives from Brisbane at 12:00pm. It departs Melbourne at 1:00pm and arrives in Adelaide at 1:40pm, with a 1 hour delay.

So instead instead of 180 passengers delayed 3 hours (540 hours), we have 180 passengers delayed 2 hours (360 hours) and another 180 delayed 1 hour (180 hours) (360 + 180 = 540).

Airline X has taken a 3 hour delay on a single flight and redistributed that single 3 hour delay across two different flights of 2 and 1 hour, respectively. By redistributing the delay they’ve managed to avoid paying compensation. The threat of compensation hasn’t removed the delay but generated an incentive for the airline to optimise their tactical plan to eliminate paying compensation. We can debates which is better or worse, however it’s entirely subjective!

EU261 was meant to reduce delays, however it inadvertently incentivises airlines to minimise compensation rather than delays. It’s reduced the delay for some passengers, but introduced a delay for passengers that otherwise wouldn’t have been delayed at all. In effect, it socialises the externalities of the delays.

A secondary and even more egregious consequence is that once a flight (with no recovery options) is delayed beyond 2 hours, the airline’s incentive switches once again. Recall that the airline pays the same compensation whether the flight is delayed 3 or 8 hours. If the airline has other flights with a risk of delay exceeding 2 hours it can manipulate operational schedules - just like we showed in the example - and pack the additional delays into the one flight that’s already due compensation. In effect, this flight now becomes the “donor flight” as the airline has an incentive to delay this flight further to eliminate >2 hour delays on other flights in the network.

For example, it can take the aircraft from the original 2+ hour delay and utilise it to avoid a 2+ hour delay elsewhere. This’ll progressively increase the delay on the donor flight. So one unlucky flight might see its delay increased from 3 hours to 8 hours to eliminate compensation due on other flights. Again, this results in concentrating delays and is objectively bad!

Gaming the system #3: overbooking

Another unintended consequence is how it gives airlines full cost transparency for overbooking, a practice where an airline sells more seats on a particular flight than there are seats available. There are several reasons why airlines overbook flights (explained below) and we’re not defending the practice of overbooking, rather considering how EU261 affects it. This is importance since the genesis of EU261 was to force airlines to pay compensation to passengers in the event of denied boarding (the term EU261 uses).

Why do airlines overbook flights?

Revenue maximisation:

Let’s assume an airline sells all the seats on a particular flight for fares between €100 and €300 each. The flight is ostensibly sold out, but on the day a customer walks up to the counter and is willing to pay €500 to be on that flight. Despite not having a seat available, the airline goes ahead and sells the customer a ticket for €500.

They’ll now have to remove a ticketed passenger from the flight to make space. If they remove someone who paid €100 they’ll make a gain of €400, while removing someone who paid €300 will result in a gain of €200. Thus, as long as the oversold ticket is more expensive than the passenger who is removed, the airline gains.

It’s even become common practice in the US for airlines to offer passengers inducements to voluntarily take a later flight. As long as this inducement costs less than €200 it’ll result in a net gain and very likely result in two happy customers!

Risk of missed connections and no shows:

Some airlines operate hub-and-spoke models that connect passengers from one flight to another. Some inbound passengers will arrive late and miss their connections. Alternatively, some passengers just don’t turn up, including those with flexible tickets. Missed connections and no-shows cost airlines lost revenue when seats go empty.

This leads them to selling more tickets with the expectation of some missed connections and no-shows. Airlines have pretty good data to estimate the number of passengers that are likely to miss a flight. In some cases, it can be quite granular with flight, day and seasonal specific estimates. The challenge comes when the number of missed connections or no-shows is less than expected, leading to oversold flights.

So now if a passenger is denied boarding due to overbooking, then the airline must pay the passenger €250 if they are unable to get the passenger to their destination on another flight that’ll arrive at their destination within 2 hours of the originally scheduled arrival time.

But here’s the kicker: now that the airline knows their penalty they also know the marginal revenue required to overbook!

In the example, the passenger that bought the overbooked ticket on the day of the flight paid €500, replacing a passenger on a cheap ticket that cost €100. The airline’s marginal revenue was €400. But the airline now knows that as long as the marginal revenue is greater than the €250 compensation then they’re in the money and a marginal profit of €150 (€400 less €250). Thus, as long as the marginal revenue exceeds compensation, the compensation generates incentives for the airline to overbook more, not less!

The problem isn’t regulation per se, but rather that airlines must pay compensation. And since compensation is fixed, and this fixed any potential cost for the airline. While the example of overbooking appears trivial in comparison to the previous two examples, it highlights how better regulation without compensation is likely to generate better outcomes. But what is this alternative regulatory option?

It’s actually quite simple, just require that a confirmed ticket guarantees a seat. i.e. an airline must seat all ticketed passengers! And if they can’t then they must find a willing volunteer to give up their seat. Willing buyer, willing seller!

Rather than fix compensation, airlines would have to offer an inducement to a passenger or several passengers to give up their seat. Instead of fixing compensation, they’ll have to pay what people are willing to accept. For example, they’ll probably have to pay a lot more to get passengers to give up their seat at 5pm on a Friday afternoon than they would at 11am on a Sunday. They’d also have to pay successively larger amounts for more passengers.

The opportunity cost for every passenger is different, and may also vary by time or day of the week. So why should the compensation be the same? Forcing an airline to find a willing volunteer ensures that overbooking is properly priced rather than being priced by a bureaucrat in Brussels or Canberra.

And if an airline ends up having to pay a lot of money, exceeding their marginal revenue, then that’s their problem. If that happens too often and costs them too much money they’ll adjust their overbooking policies accordingly. Ultimately, it’ll force airlines to price risk accordingly, rather than fixing their risk based on an arbitrary price.

Concluding thoughts

No doubt, some will read this and label us as (airline) apologists. Australian airlines have seen their reputations go downhill in recent years and they’ve become media cannon fodder. High rates of cancelled and delayed flights have added fuel to the fire, especially in the aftermath of the COVID-19 pandemic. We recognise the displeasure the public has with the industry and how this makes compensation policies like EU261 popular. Such policies will penalise the perpetrators and give the aggrieved a sense of justice. We get it and we understand it!

But what is the purpose of compensation policies like EU261? Is it an archaic sense of justice where perpetrators of bad things must face penalties to bring about justice for the victims? Let’s be frank, this isn’t a violent crime or property crime. It’s a cancelled or delayed flight.

Instead of being apologists, we’re simply showing how compensation policies aren’t likely to be effective at reducing flight cancellations or delays, or even reducing the length of those delays. Airlines are cunning and they’ll respond to the incentives in front of them. The challenge for regulators is that compensation policies generate incentives for airlines to minimise compensation, not minimise cancellations and delays.

We showed some mechanisms they’ll use to game the system, for example changing the distribution of delays rather than the total quantity of delays. So instead of only some passengers being delayed, a larger number of passengers who would otherwise not have been delayed will now have to be delayed in order to minimise compensation. Alternatively, passengers who are due compensation may experience longer delays than they otherwise would have to limit compensation having to be paid to other passengers. We also showed how the application of compensation to overbooking might actually increase overbooking as fixed compensation provides transparency of the cost of overbooking for airlines.

So what should regulators do? It’s rich to critique what they shouldn’t do without actually saying what they should do! Firstly, policy makers would be better advised to conduct a thorough and comprehensive assessments regarding the cause of cancellations and delays. This seems like we’re punting, but without understanding the cause and variance of cancellations and delays, it’s difficult to design solutions to reduce them.

Secondly, what about introducing more graduations into the system, as suggested during discussions on social media? There’s some validity to this, but this introduces even more (arbitrary) discrete thresholds with fixed cost parameters for airlines to optimise around. It misses the point that a compensation based systems will always have an objective function to minimise penalties, not improve underlying behaviours.

Thirdly, require transparency. Some cancellations and delays are inevitable or unavoidable, whether they’re a result of mechanical faults or poor weather. Others aren’t inevitable or unavoidable and the responsibility of airlines, whether that be due to poor planning or execution. However, we very little understanding of cancellations and delays in Australia.

For example, BITRE data on cancellations and delays just tells us the percentage of flights cancelled or delays by airline and route. Inexplicably, it doesn’t tell us anything about the length of delay and a 16 minute delay is counted the same as an 8 hour delay. It also doesn’t tell us about the day of the week but more importantly it says nothing about cause of delay and thus who bares responsibility! This piles the responsibility on airlines and absolves the rest of the supply chain of responsibility, whether that be weather, airport congestion or ATC staffing shortages.

The other side of transparency is not about regulation, but airlines themselves. Most airlines don’t have a culture of transparency when it comes to cancellations and delays. Often they don’t say anything while at other times they’ll say vague things like "your flight is delayed by weather".

Why don't more they communicate more transparently? Do they think people don't want to know, don’t care or can't understand? We’re not suggesting a sinister motive, rather just risk aversion, but we argue that transparency builds trust!

Since 2018, United Airlines have been taking a very different approach with a narrative of "every flight has a story". They argue that it reinforces a culture of transparency and builds trust. In many cases, they’re providing mundane details but it pushes back against the commoditisation of air travel, just a little.

It's the mundaneness that is so brilliant! It would be very easy to just come across with sophistry (clearly why we shouldn't be allowed to write them!!!). It's got to be specific enough not to seem generic and vague, but mundane enough to avoid appearance of sophistry!

Transparency won’t solve the problem, but it also shows that arbitrary thresholds and penalties will do the same. It’ll give a handful of people a good story to tell their mates at the pub about how they got a few hundred bucks off Virgin or Qantas for a delayed flight, while all their other mates complain they didn’t get compensation despite their flights being delayed.

Meanwhile, the number of cancelled and delayed flights hasn’t declined, and we can’t even say if the length of delays has changed, or whether there have been any changes in the reasons for those delays.

But it would’ve created a whole new bureaucracy to implement it, including dodgy ambulance chasers offering their help getting compensation. Policy makers will pat themselves on the back for implementing policies to help the public, despite it not really doing so. Welcome to AU261!

A very common comment that you’ll come across when engaging in the popular aviation discourse is that route X to Y will be a good size for **insert type of regional jet**. The context will be in relation to a thinner regional route, for example when Bonza went bankrupt in April 2024, many comments were that many/most of their routes would have been better suited to smaller aircraft like a 100 seat E190 as opposed to the 186 seat B737-8 aircraft they were flying.

The comment isn’t misplaced as many/most of Bonza’s routes didn’t have sufficient demand to fill 186 seats at a price that was able to cover their costs and earn a sufficient profit to generate the required return on assets. This is important as Bonza’s administrators specifically pointed to their pricing as one of the reasons for their failure. The administrators calculated Bonza's average fare at A$ 104, 26% lower than market average of A$ 141.

Essentially, in order to fill aircraft sufficiently, Bonza were having to price tickets too low to cover their costs. The challenge is that had Bonza raised their prices they likely would have seen lower passenger numbers. Higher prices would have resulted in some customers switching to alternatives like other airlines, connecting options or even driving. Furthermore, lower prices were stimulating demand, and thus higher prices would have failed to stimulate demand.

So smaller aircraft would have been more sustainable, however there is a key problem as the smaller the aircraft, the higher the unit cost. Therein lies one of the most misunderstood dynamics in the business: the cost dynamics of regional jets. Let’s explore …

Some context about cost structure

There are two important definitions that we need to understand for this analysis:

• Trip cost: the total cost of making a given trip.

• Unit cost: the total cost of making a given trip divided by the number of passengers the aircraft carries, often presented as the cost per available seat kilometer (CASK) or mile (CASM).

As a generalisation, smaller regional jets like an E190 will have a lower trip cost than a narrowbody jet like a B737 or A320 for a given trip as it’ll burn less fuel, incur lower enroute navigation charges, lower airport fees, lower staff costs (fewer cabin crew), etc.

However, smaller regional jets will have a higher unit cost than narrowbody jet on that same route, as the larger narrowbody jet will be able to share the fixed costs over a larger number of passengers and because variable costs increase at a decreasing rate (declining marginal costs).

Conceptually, this makes sense, but is there empirical data to support this? Data from the United States allows us to show this in action. We use a single airline (Delta Airlines) to show the relationship between unit cost and aircraft size, and trip cost and aircraft size across their regional and narrowbody fleet fleet. Using a single airline is import since we can assume a consistent underlying cost structure.

Using this example, the first figure (below) shows us how the unit cost of Delta’s fleet declined as as the aircraft size increases. There is one outlier at the 110 seat size, however the data shows a clear picture. For example, the B737-800s has a unit costs of 7.1 cents per available seat mile, 24% lower than the 9.3 cents per available seat mile of the CRJ 900.

However, the second figure shows how trip cost increases with aircraft size, with the CRJ 900 - despite the higher unit cost - has a significantly lower trip cost than the B737-800. The CRJ 900 will cost US$ 2686 to cover a 500 mile trip, 53% less than the US$ 5671 on the B737-800.

So the empirical evidence supports the conceptual argument that larger narrobodies are more efficient than their smaller regional jet contemporaries. But there’s a big but, as this only benefits the airline if they are able to fill the extra space, and fill it at a higher enough price. However, flying around larger aircraft generates more risk due to the higher trip cost.

Meanwhile, a regional jet presents lower risk than a narrobody jet since its trip cost is much lower in the event of an empty flight. However, its higher unit cost means that it’s required to earn a higher unit revenue (revenue per available seat kilometer or RASK), all else being held constant. This presents a conundrum for airlines.

So why do airlines then operate regional jets?

The contemporary history of regional jets takes us back to the US where major legacy airlines like American, Delta and United have an extensive fleet of regional jets flying with feeder airlines. These typically operate under linked but seperate brands (e.g. American Eagle, Delta Connection and United Express). The feeder airlines are wholly owned subsidiaries (e.g. Delta’s subsidiary Endeavor Air) or contracted out to independent firms (e.g. Delta also utilise capacity operated by Republic Airways and SkyWest Airlines).

All US legacy carriers are strongly dependent on the hub-and-spoke model due to the fragmentation of the US population with most hubs operating as hybrids between banked and rolling hubs to differing extents. Irrespective of the banking strategy, hub-and-spoke models favour higher frequency operations, lending itself to more flights on smaller aircraft rather than fewer flights on larger aircraft, all else being held constant.

For example, if an airline has two aircraft types, a 50 seat regional jet and a 150 seat narrowbody, a route with an expected daily demand for 150 seats can be operated 3x daily on the 50 seat regional jet or 1x daily on the 150 seat narrowbody. The higher frequency operation will enable more connections and offer more flexibility for customers, thereby increasing yields and unit revenue. However, the million dollar question is whether the increase in yields will be sufficient to increase unit revenue more than the increase in unit cost imposed by regional jets?!

The second consideration is that crew at regional carriers are paid less than their counterparts at mainline carriers. This led to the development of “scope clauses” in labor bargaining agreements that limit the capacity that mainline carriers can employ with regional jets. For example, Delta’s agreements with crew unions limits them to operate a maximum of 125 aircraft with 50 or fewer seats, 102 aircraft with 51-70 seats, and 233 aircraft with 71-76 seats (there are other constraints that limit stage lengths and hub versus non-hub flying).

Another interesting artefact of US regional jets is how some operate with relatively low density premium layouts. This is ostensibly to maintain aircraft within scope clause limits. For example, Delta’s CRJ 900s are fitted with 76 seats (12 first/business and 64 economy) whereas many other legacy carriers like Lufthansa and SAS operate the same aircraft with 90 seats. Other examples include the Delta’s E175 with 76 seats (12 first/business and 64 economy) compared to 88 and 82 on KLM and LOT, respectively.

The same is found at other cutoffs, sometimes with rather peculiar configurations. For example, United wanted to replace some 50 seater regional jets, but the with the CRJ 200 out of production, they took new CRJ 700s with a reduced MTOW and fitted it with 50 seats (10 first/business and 40 economy), calling it the CRJ 550. Comparatively, United’s original CRJ 700s seat 70 passengers (6 first/business and 64 economy). All this to remain within the scope of the scope clauses!

It works for US legacies, so why isn’t this a generalisable solution?

Regional jets are a generalisable solution, as evidenced by them operating in many other countries, not just the US. Scope clauses aren’t just a US phenomenon either with them being present in Canada, while other countries may have similar considerations within Enterprise Bargaining Agreements, although fixed limits on the number of aircraft and aircraft size isn’t as widespread. However, the complication remain the same: can the airline generate sufficient revenue premium to offset the higher unit cost?

A great example of this is Qantas who utilise a range of regional jets under the QantasLink brand. In some cases they use regional hets to operate a single daily flight on thinner routes (e.g. Qantas deploy the A220 on its once daily Melbourne-Coffs Habour flight), while on other routes it allows them to operate at greater frequency than they would utilising a narrowbody (e.g. Qantas deplot the A220 4x daily on Melbourne-Hobart), thus generating more connections and increasing passenger flexibility.

But it means that the higher unit costs needs to be recovered in higher ticket prices. On thiner routes, it results in higher ticket prices for all passengers while on routes where regional jets are deployed to increase frequency it can be recovered through a yield premium for some passengers on some flights through price discrimination including generating higher yields off increased network flows and passengers willing/able to pay for greater schedule flexibility.

But this isn’t generalisable to all airlines. For example, this simply doesn’t work for most LCCs as their business model relies on reducing or minimising unit costs relative to competing or incumbent carriers. This is needed to allow for lower unit revenues and thus ticket prices. In fact, as we’ve argued before, LCCs must reduce unit costs by more than the decline unit revenues to achieve the same return on the asset.

The lesson: the higher unit costs of regional jets doesn’t suit LCCs

So back to Bonza, it’s simply not the case that their routes were better suited to a regional jet, well at least not within their business model! While regional jets would better match capacity with demand, it would significantly increase unit costs and thus require them to increase ticket prices, which as we’ve already noted, were too low to cover the costs of the B737-8s.

But this isn’t really Bonza specific. Just like any LCCs, they wouldn’t be able to generating higher yields from increased network flows and passengers willing/able to pay for greater schedule flexibility, since they didn’t have this to begin with. It simply isn’t part of the business model, or any conventional LCC for that matter.

But this also goes a long way to explain why LCCs typically don’t fly regional jets. It’s often argued that LCCs typically operate one fleet type as a cost savings measure for crew and maintenance. This is true, but it’s also a function of the narrower scope of aircraft that suit the LCC business model. Regional jets typically don’t suit their business model due to higher unit costs, while larger widebody aircraft undermine other cost structure needs, including turnaround times.

Smaller niche LCCs might find the lower trip cost useful, particularly in environments with limited or not competition, or where incumbents are so inefficient that even flying regional jets would result in a lower unit cost than incumbents, but these sound more like outliers than generalisations.

Our last chart of the week looked at the most recent Australia-Europe market shares, showing Emirates and Qatar Airways to be the clear market leaders with 30% and 21% market share, respectively, and Singapore Airlines (11%) and Qantas (9%) a distant 3rd and 4th.

The context of that post was the return of Virgin Australia Qatar Airways to long haul flying with their inaugural Sydney-Doha flight. The post highlighting how Emirates are likely to see their dominance between Australia and Europe significantly challenged in the coming years.

A more interesting proposition is how these market shares have changed over time, and particularly how the rise in Gulf carriers like Emirates, Etihad Airways and Qatar has affected home market carriers like Qantas and British Airways, and transit carriers like Singapore Airlines.

It’s a more complicated question than it appears. And looking at the evolution of this traffic over time we also thought that it’s better to separate the United Kingdom from Europe, but why?

ABS data shows that 43% of travel between Australia and Europe over the last year (year ending June 2025) was to/from the UK and Ireland. By comparison the next largest are Italy, France and Germany with 11%, 8% and 8%, respectively, and the remaining 30% of the market is highly fragmented. In many cases, these markets aren’t available as one-stop connections on many carriers, whereas Gulf carriers offer a much wider range of European one-stop connecting options to/from Australia. However, one thing all these carriers do have in common though is that they all serve the UK!

Number of one-stop destinations served between Australia and Europe:

Qatar Airways: 50
Emirates: 32
Etihad Airways: 37
Cathay Pacific: 12
Singapore Airways: 12
Qantas: 3
Malaysia Airlines: 2
British Airways: 1

A second complication is that while transit carriers offer their service through a single hub (i.e., Emirates through Dubai, or Malaysia through Kuala Lumpur), home carriers like Qantas and BA (and historically Virgin Atlantic) have offered their services through a range of hubs, or in the case of Qantas, direct on the Perth-London route.

While Qantas and BA currently push all their transit traffic through Singapore, Qantas previously transited through Bangkok, Dubai and Hong Kong, while BA have also operated via Bangkok, and when Virgin Atlantic operated the route they transited via Hong Kong.

So to answer the question, we focus the analysis on the UK rather than Europe, and airlines rather than transit points!

We didn’t have to do much original data analysis on this, rather we looked through the treasure trove of ACCC submissions, going as far back as Qantas and BA’s joint venture application (2000), Qantas and Emirates’s applications (2013, 2018 and 2023), and most recently Virgin and Qatar’s (2024). This allowed us to piece together the market shares, albeit with some breaks and inconsistent time periods. It paints a fascinating picture!

Looking back at the 1990s, the Gulf carriers played no role in the Australia-UK market! This isn’t surprising as Emirates only entered the Australian market in 1996. Other gulf carriers like Gulf Air served Australia but on a much smaller scale, and Etihad and Qatar only entered the market in 2007 and 2009, respectively.

In the late 1990s, Qantas and BA were the dominant carriers between Australia and the UK with 31% and 14% of the market each. Qantas offered 2x/day flights, while BA offered 3x/day flights. By 2003, Qantas had increased their offering to 3x/day (and a 4th in late 2004), and saw their market share increase to 34%, while BA saw their market share decline to 10%.

Qantas’s market share consistently declined, falling to 25% in 2012 before cratering to 16% in 2014 following the reduction in their number of flights from 4x to 2x/day in 2012. Their market share has since recovered to 20% following the introduction of non-stop flights from Perth. Meanwhile, BA’s market share is now just 5%, following a reduction to 2x/day in 2006 and to its present offering of 1x/day in 2012.

Qantas and BA’s loss has been Emirates’s gain! From nothing in 1998 to 32% in 2014, Emirates has become a juggernaut! Since it’s 2014 peak, Emirates’s market share has declined to 23% of the market, however they remain the market leader. Interesting, Emirates’s declining market share over the last decade coincided with the entry of Etihad and Qatar, although Etihad’s own troubles have reduced their market share in recent years.

Prior to the COVID-19 pandemic, Emirates held 26% of the market in 2019, ahead of Etihad and Qatar with 11% each, combining for 48% of the market. This suggests that a significant chunk of Etihad and Qatar’s growth was simply cannibalising Emirates, as nearly a third of Etihad and Qatar’s gain since 2019 was accounted for in Emirates’s decline. Notably, Qantas, Singapore, Cathay and BA held their market shares during this time, indicating that smaller carriers like Malaysia were also significant casualties (7% to 2%).

Since the pandemic, Emirates has lost further ground, seeing their market share decline to 23%, while Etihad have wilted to just 4%. The combined loss of 10 percentage points has been offset by gains at Qatar and Qantas, gaining 6 and 4 percentage points each.

One constant player over the last decade has been Singapore, maintaining a market share of between 11% and 13% since 2010. However, they have also felt the impact of Gulf carriers, having consistently achieved market shares between 16% and 18% in the early 2000s. Cathay also maintained relatively consistent market shares over time, but have struggled to return capacity in the post pandemic era.

Conclusion

Over the last two decades, Gulf carriers have come to dominate the Australia-UK market, now accounting for nearly half of the market. Qantas and BA have been the major casualties, although Qantas have done well to recover and increase market shares in recent years. Smaller transit carriers like Malaysia Airlines have suffered large losses, while larger transit carriers like Singapore have also suffered losses.

The more recent emergence of Qatar has thrown a spanner in the works as they now compete effectively with other Gulf carriers, winning significant chunks of market share from Emirates and Etihad. No carriers have been safe and to some extent, Qantas and Singapore remain ever present, having held their market share well over the last decade while the Gulf carriers fight it out, and in the case of Qantas, have even seen meaningful gains.