Three Businesses, One P&L: Nights Accelerate and EBITDA Sets a Record, But the Take Rate Slips and the 2026 Margin Answer Is Deferred to February

Results vs. Consensus

Q3 2025 Scorecard

Year-Over-Year Comparisons

Quarter-Over-Quarter Comparisons

The sequential table is the least informative view of this business and needs a caveat rather than an interpretation. Airbnb books GBV when the reservation is made and recognizes revenue when the guest checks in. Q2 is the peak booking quarter for northern-hemisphere summer travel; Q3 is the peak stay quarter. That timing mismatch is why GBV can fall 2.6% sequentially while revenue rises 32.3%, and why the implied take rate swings from 13.2% to 17.9% on no change whatsoever in fee structure. Year-over-year is the only comparison that carries signal for this company, and we treat it that way throughout.

Quality of the Print

Revenue: The 9.7% growth rate is the sixth consecutive quarter in which Airbnb has grown revenue in the 6–13% band, and it is the same figure ex-FX. What is new is the composition. Nights contributed 8.8 points of the 13.9% GBV growth and ADR the remaining 4.7, but ADR growth was only 2% ex-FX, meaning currency supplied roughly half of the price contribution. Strip currency out of GBV entirely and the underlying figure is +12%, still a two-point acceleration from Q2 but a smaller one than the headline advertises. The honest read is that volume improved genuinely and price improved partly on translation.

Margins: The 118bp of operating margin compression decomposes cleanly and the components sum. Cost of revenue added 95bp of pressure (up 18.1% against 9.7% revenue growth, tracking payment-processing intensity as installments in Brazil and Reserve Now, Pay Later in the US scale). Sales and marketing added 183bp (up 24.3%, the single largest cost delta in the quarter). Product development added 29bp. Against that, operations and support gave back 98bp and general and administrative 92bp. Note what operations and support did: it declined in absolute dollars, down 1.1% year-over-year, in a quarter when nights grew 8.8%. That is the AI customer-support deployment showing up in the P&L as real operating leverage, roughly a quarter ahead of when we would have expected it. The problem is that the leverage is being spent on customer acquisition faster than it is being earned.

EPS: The path from +6.6% operating income growth to +3.8% diluted EPS growth runs through three below-the-line items and one offset. Interest income fell 13.0% to $180M as the rate cycle turned against a company sitting on $11.7B of corporate cash. Other income swung to a $13M expense from $3M of income. The tax rate rose 215bp on the CAMT allowance. Working the other way, the buyback took 21M shares out of the diluted weighted average, a 3.3% reduction worth roughly 3.3 points of EPS growth on its own. Absent the buyback, EPS growth would have been roughly flat at about +0.5%, so the repurchase supplied essentially all of the reported growth. That is not a criticism of the buyback, which is the correct use of $4.5B of trailing free cash flow, but it does mean the reported EPS line is currently doing more work than the P&L beneath it.

Key Business Metrics

Airbnb does not report operating segments. The disclosure that substitutes for segment reporting is the marketplace KPI set (nights, GBV, ADR, implied take rate) plus qualitative regional growth bands. We treat the KPI set as the primary analytical surface and the regional bands as the segment view.

Regional Performance

North America — the swing factor, and the reason to read the footnote

North America is roughly 30% of total Nights and Seats Booked and was the source of the quarter's acceleration. It is also where the acceleration has an identifiable mechanical driver. Reserve Now, Pay Later launched in the US in August, letting guests book eligible domestic stays with $0 upfront. The CFO disclosed that of the guests to whom the option is offered, about 70% take it. That is an extraordinary attach rate, and it is exactly the kind of number that should make an analyst want to know the cancellation offset before crediting the growth.

"So yes, there are increased cancellations, but we're highly confident that the net impact of the product is a lift to the net bookings." — Ellie Mertz, CFO

Assessment: We accept that the net effect is positive; the company tested the product before launching it and would not scale a feature that destroyed net bookings. The question is not sign but magnitude and durability. Nights and Seats Booked is reported net of cancellations that occur in the booking quarter, and management explicitly flagged that Reserve Now, Pay Later cancellations will skew toward dates closer to the stay, which "may or may not occur in the same booking quarter." That means some portion of Q3's reported nights is exposed to cancellation in Q4. It also means the year-over-year lift from the feature is a one-time step, not a growth rate, and it laps in Q3 2026. We would not extrapolate the US acceleration.

Latin America and Asia Pacific — the expansion thesis is working

Latin America at low-20s nights growth and Asia Pacific at mid-teens are running at roughly two to three times the company average, and management disclosed that expansion-market nights on an origin basis grew at double the rate of core markets over the trailing twelve months. The country-level evidence is specific rather than promotional: Japan domestic nights +27% with three consecutive quarters of acceleration, India first-time bookers up nearly 50%, Brazil first-time bookers +17% with the share of bookers choosing an installment option more than doubling after the July launch.

Assessment: This is the most credible growth vector in the story and it is under-discussed relative to the AI narrative. The playbook is unglamorous and repeatable: localize product, localize payments, localize marketing, wait three years. Brazil is the proof case and Japan is the current cohort. The constraint is that these markets carry structurally lower ADR than North America and EMEA, so mix shift toward them is a slow drag on blended ADR and take rate even when it is unambiguously good for volume and for the long-run TAM.

EMEA — the comp, not the trend

Mid-single-digit nights growth in EMEA against a Paris Olympics comp is a better result than it looks, and the 10% ADR gain is mostly translation (only +4% ex-FX). We would treat EMEA as running roughly in line with North America on an underlying basis and would expect the comp to normalize in Q3 2026.

Assessment: No thesis content here in either direction. EMEA is the mature half of the business and behaves like it.

Supply

Active listings grew approximately in line with Nights and Seats Booked, which is the balance a marketplace wants: supply keeping pace with demand rather than running ahead (which depresses host economics) or behind (which depresses conversion and raises price). Growth skewed toward Latin America and Asia Pacific, matching where demand grew fastest. On quality, management disclosed that more than 550,000 listings have been removed since the 2023 hosting-quality system launched, and that customer-service issue rates and chargeback rates both declined year-over-year in the quarter. Guest Favorites has now accumulated nearly 500 million nights booked.

Assessment: Supply is not the binding constraint on this business and has not been for some time. The binding constraints are price competitiveness versus hotels and the quality variance that keeps roughly nine of every ten lodging nights in a hotel. Management is addressing both, but slowly, and neither is a 2026 event.

Key Topics & Management Commentary

Overall Management Tone: Confident on product and deliberately vague on economics. The prepared remarks and the Q&A ran heavily toward what is being built (65 product improvements, three businesses, more than a dozen AI workstreams) and lightly toward what any of it costs or earns, with the CEO carrying most of the call and volunteering an unusually long unprompted AI answer in anticipation of the question. Where management was least convincing was the one place it mattered most: asked directly whether the 2025 investment in new businesses rolls off in 2026, the answer was a directional assurance with no numbers attached and an explicit deferral to the February call.

1. Nights Accelerated Two Points, and a Payments Feature Is Part of Why

Nights and Seats Booked grew 8.8% to 133.6M, up from roughly 7% in Q2, with the acceleration concentrated in the United States. Management was direct that Reserve Now, Pay Later was a contributor rather than leaving it to inference, and the mechanics are worth understanding because they determine whether this is a level shift or a growth rate.

"Nights and seats booked increased 9% year-over-year, representing a 2-point sequential acceleration from Q2, primarily due to the strength in the U.S. In Q3, nights growth across each of our major regions remained steady or accelerated sequentially." — Ellie Mertz, CFO

The feature is offered only to US guests travelling domestically who select listings with flexible or moderate cancellation policies, so it is not yet available to the whole US base. Within that eligible population the attach rate is roughly 70%. Management plans to roll it out more broadly next year and has flagged additional payment instruments as a growth lever, pointing to Brazilian installments as the template.

Assessment: Constructive but bounded. Reducing the cash barrier at the point of booking raises conversion in a business where the decision is discretionary and the ticket is large; the Brazil installment result (share of bookers choosing installments more than doubled) is independent confirmation that the mechanism works. What it is not is a compounding driver. Each payment innovation delivers a one-time conversion step in the geography where it launches, then laps. We model the US contribution as fading through H2 2026 and would look to international rollout to replace it rather than add to it.

2. The Take Rate Fell 69bp, and October's Fee Restructure Complicates the Comparison

Implied take rate, which the company defines as revenue divided by GBV, was 17.9% against 18.6% a year ago. Management attributed the decline to two factors, neither of which is a pricing concession.

"primarily due to FX and the timing of when guests booked their travel and when guests stayed." — Q3 2025 Shareholder Letter

Separately, and after quarter-end, Airbnb changed its fee architecture. In October the company migrated property-management-software hosts off the split structure, under which hosts paid a 3% fee and guests paid a separate service fee, onto a single 15.5% host fee. Non-PMS hosts already on the single-fee structure will see the same 15.5% rate applied. Management's framing is that this lets hosts price more competitively, since a host can lower the displayed nightly rate and keep the same net earnings.

Assessment: Two separate issues are being conflated by the market and should not be. The 69bp of Q3 compression is largely mechanical and should partially reverse. The October fee migration is the one that matters, and it is deliberately deflationary to the displayed price: management is trading a smaller monetization wedge on each booking for a lower shelf price and, it hopes, more bookings. That is the right long-run trade against hotels, and it is exactly the trade that makes the next several quarters of take rate hard to read. Management guided Q4 implied take rate to be roughly flat year-over-year, which suggests the fee change is close to neutral in aggregate in its first quarter, but we would treat take rate as an unreliable near-term signal and watch GBV and revenue growth separately.

3. The $213M CAMT Charge, and a Structurally Lower Tax Rate from 2026

The quarter's tax provision of $418M on $1,792M of pre-tax income implies a 23.3% effective rate, up from 21.2%. The step-up is entirely attributable to a one-time valuation allowance recorded against deferred tax assets tied to Corporate Alternative Minimum Tax credits, triggered by the July 4 enactment of the One Big Beautiful Bill Act.

"Now net income was impacted by a onetime $213 million valuation allowance related to corporate alternative minimum tax credits as a result of the enactment of the One Big Beautiful Bill Act on July 4. On a go-forward basis, starting in 2026, we anticipate that the One Big Beautiful Bill will materially reduce our effective tax rate due to the preferential changes to tax on foreign earnings." — Ellie Mertz, CFO

The company now expects a full-year 2025 effective rate of approximately 21%, reflecting the $213M charge partially offset by a one-time $60M release of an uncertain tax position and an ongoing foreign-derived intangible income benefit. From 2026 the expected long-term rate falls to the mid-to-high teens.

Assessment: The single most under-priced item in the print, and it runs in the shareholders' favour. A move from roughly 21% to, say, 17% on a pre-tax base of roughly $3.4B is worth on the order of $136M of net income, or roughly $0.21 per diluted share, arriving with no operating effort at all. The market spent the evening reacting to a $0.34 charge and paying comparatively little attention to the recurring benefit that follows it. We build the mid-to-high-teens rate into our FY26 estimates and flag it as the most reliable EPS tailwind in the model.

4. Experiences and Services: Real Signal, Distant Revenue

Six months after the May relaunch, the evidence on experiences and services is better than we expected on engagement and exactly as expected on monetization, which is to say there is not any yet. Guest ratings average 4.93 out of 5. Roughly half of experiences bookings in Q3 carried no associated accommodation booking, and management said that for services specifically, 10% of bookers had never transacted on Airbnb before. Host-side supply interest is scaling faster than either: more than 110,000 applications from prospective services and experiences hosts, which the CEO said had nearly doubled from the prior quarter.

The company is deliberately not scaling this globally. It is running Paris and Los Angeles as depth markets to establish product-market fit before extending the playbook to roughly a dozen more cities, then hundreds.

"So it's going to take 3 to 5 years, I think, for service experiences to become a material part of our business, but I'm very, very bullish on them." — Brian Chesky, Co-Founder & CEO

Assessment: Credit for the honesty of the three-to-five-year framing; it is more useful than a TAM slide. But investors should hold the two halves of this together. The engagement data is genuinely encouraging and supports the argument that adjacencies widen the funnel rather than cannibalise it. The financial reality is that this line consumes roughly $200M of 2025 operating expense and contributes no disclosed revenue, and by management's own timetable will not be material before 2028. For a stock trading on 2026 free cash flow, experiences is an option with a long expiry, and options with long expiries are worth less than the enthusiasm they generate.

5. Hotels: The Cheapest New Business Airbnb Has Ever Launched

Airbnb began piloting a purpose-built hotels product in Los Angeles, New York City and Madrid, partnering directly with boutique and independent properties. The product work is substantive rather than cosmetic: new search filters, redesigned hotel page displays, and room-type selection. Two of the three pilot markets, New York and Madrid, are markets where short-term-rental regulation constrains Airbnb's core supply.

"And so to add hotels is not really an incremental investment at all. And to recruit hotels is actually much easier than recruiting homes because a hotel, they have a phone number, you can call them. Homes, they have to primarily come organically." — Brian Chesky, Co-Founder & CEO

The strategic logic is demand-side arbitrage. Airbnb reaches roughly 1.6 billion devices a year and generates enormous search volume in exactly the urban markets where its home supply is regulatorily capped. Historically that traffic leaked to hotel booking apps. Hotels convert leaked traffic into commission revenue with no incremental demand-generation spend.

Assessment: The most economically attractive of the three new businesses and the one closest to revenue, precisely because it inverts the marketplace cold-start problem. Airbnb has the demand and must acquire supply; supply here is a phone call to a general manager rather than a multi-year organic host-acquisition machine. The honest risks are two: boutique and independent inventory is a fraction of global hotel room-nights and will not move the revenue needle at pilot scale, and stepping onto the incumbent online travel agencies' turf invites a competitive response in a channel where those incumbents spend far more on performance marketing than Airbnb does. We would not model hotel revenue in 2026 but would treat a pilot-to-scale announcement as a genuine upside catalyst.

6. AI Is Already in the P&L, Just Not Where the Narrative Says

The CEO gave the longest single answer of the call on AI, framing a three-phase strategy: customer support first (deliberately chosen as the hardest problem, since hallucination is unacceptable and the data is sensitive), then AI-powered search, then a unified assistant spanning discovery, booking, messaging and post-trip.

"Now we initially launched this in the United States, where it's already reduced people's need to contact a human agent by 15%. So now we're going to expand it to more countries in more language, and we expect this to be in over 50 languages next year." — Brian Chesky, Co-Founder & CEO

The competitive framing was that model access is not the moat, because every frontier model is available by API to every participant.

"And our thesis of AI is that specialization will win in travel. That's our theory, that specialization will win." — Brian Chesky, Co-Founder & CEO

Assessment: The AI item worth owning in the model is not conversational search; it is the operations and support line, which fell 1.1% in absolute dollars while nights grew 8.8%. That is 98bp of cost-ratio relief in a single quarter, and with the assistant expanding to more than 50 languages next year, the leverage should compound through 2026. The search and assistant work is real but unquantifiable today and carries a genuine strategic risk the company has not addressed: if travel discovery migrates to general-purpose assistants, "specialization will win" is a hypothesis, not a moat. We give credit for the support-cost leverage now and none for the search roadmap until it converts.

7. The ChatGPT Absence Was a Choice, and It Is a Real Strategic Bet

Airbnb was conspicuously missing from OpenAI's app-integration launch while other major travel players participated. Management addressed it directly rather than deflecting, giving four explicit principles: launch partnerships require custom integrations, the company will not appear as a commodity, it will not be a data layer, and users must be able to see what makes Airbnb listings distinctive.

"We didn't want to be positioned as essentially a data layer like a commodity." — Brian Chesky, Co-Founder & CEO

The precedent cited was Google Hotel Finder, which Airbnb also declined, on the grounds that presenting Airbnb listings alongside hotels trains users to treat the two as interchangeable.

Assessment: Internally consistent, defensible, and expensive if wrong. The brand-integrity logic is sound and the Google Hotel Finder precedent shows this is a standing policy rather than a reaction. But the position depends on Airbnb's 90%-direct-and-unpaid traffic mix holding while a new discovery surface forms. If assistant-mediated booking scales faster than Airbnb's own AI search, the company will have ceded the top of a new funnel to preserve presentation control on the old one. Management left the door open ("I would imagine in the future that you would see Airbnb across a large surface area of the Internet"), which is the right posture. We flag it as a live risk rather than a resolved one.

8. International Expansion: The Quietest Part of the Story and the Most Proven

Expansion-market nights on an origin basis have grown at roughly double the rate of core markets over the trailing twelve months. The company frames the strategy as multi-year and market-by-market rather than a global push: localize the product, localize the marketing, localize the payment instruments, then wait.

Brazil is the mature case, entered several years ago, now delivering both meaningful share and continued annual share gains. Japan is the current cohort, roughly a year into the local-marketing effort, with domestic nights +27% and three consecutive quarters of acceleration. India is earlier still, with first-time bookers up nearly 50%.

Assessment: This is where we would put incremental capital if we were running the company, and it is the vector we would credit most heavily in a bull case. It has a demonstrated template, a measurable leading indicator (first-time bookers), and a multi-year runway across a business where five countries still account for the majority of revenue. The offset is that expansion markets carry lower ADR, so success here shows up as volume and TAM rather than as near-term revenue or take rate. Investors looking for the acceleration to appear in the reported revenue line within four quarters will be disappointed by a strategy that is working.

9. Margins: FY25 Raised to ~35%, FY26 Explicitly Not Framed

Management raised the full-year 2025 adjusted EBITDA margin outlook to approximately 35%, up from a previously communicated floor of 34.5%. Against FY24's 36.4% that is still roughly 140bp of full-year compression, and the Q4 framing is for adjusted EBITDA to be flat to down slightly year-over-year with margin declining, driven by investments in new growth and policy initiatives.

On 2026, the shareholder letter said the company is "focused on maintaining strong margins while continuing to invest in growth initiatives." Pressed in Q&A on whether the roughly $200M of 2025 investment in services and experiences rolls off, the CFO gave a directional answer and deferred the number.

"But where we sit today, we feel like we can afford those investments while continuing to deliver great core efficiencies from our core business and in doing so, deliver net-net a continued strength of margin. So more color on that to come." — Ellie Mertz, CFO

Assessment: "Continued strength of margin" is a phrase that survives being wrong. It is compatible with 35% and with 33%, and the difference between those two on roughly $13.3B of 2026 revenue is about $270M of EBITDA. The substantive content in the answer was that 2025 carried first-year launch heaviness that will not repeat, but that experiences, services, hotels and AI will all be funded in 2026. That is four investment lines where 2025 had two. We assume flat-to-slightly-down 2026 adjusted EBITDA margin versus FY25's ~35% and treat any February guide at or above 35% as a positive surprise.

10. Capital Allocation: Three Quarters of Free Cash Flow Returned

Airbnb repurchased $857M of Class A stock in the quarter and more than $3.5B over the trailing twelve months.

"And during the trailing 12 months, we have repurchased over $3.5 billion of our common stock, using approximately 3/4 of our free cash flow." — Ellie Mertz, CFO

The remaining authorization stands at $6.6B, including the $6B added in Q2. Fully diluted share count fell to 646M from 665M a year ago, and management reiterated the cumulative reduction of approximately 8% since the program began in 2022. The company ended the quarter with $11.7B of corporate cash and investments against $2.0B of debt now classified as current, plus $7.2B of funds held on behalf of guests.

Assessment: Unambiguously the right policy and executed at the right price. Buying back roughly 3% of the diluted count annually at 15x EV to free cash flow, funded entirely from operations with $9.7B of net cash still on the balance sheet, is accretive arithmetic. Two caveats. First, gross repurchases substantially exceed net share-count reduction because stock-based compensation runs at $399M a quarter, roughly 9.7% of revenue; a meaningful share of the buyback is offsetting dilution rather than shrinking the base. Second, the $2.0B of debt moving to current means a 2026 refinancing or repayment decision that will consume some of the cash balance. Neither changes the conclusion, but both temper how much of the buyback should be counted as shareholder return.

11. Product Velocity as the Stated Strategy

Sixty-five major product improvements shipped since May, including updated maps with landmarks and transit, revised cancellation policies allowing free cancellation up to 14 days before check-in, and flexible location carousels that surface listings just outside the guest's stated search. Management was explicit that there is no single unlock in this business.

"there's not a lot of silver bullets in our business. It's going to be hundreds and hundreds and hundreds of improvements that you're going to see over the coming years." — Brian Chesky, Co-Founder & CEO

The CFO extended the point to the two areas where the roadmap is largest, quality and affordability, framing both as multi-year rather than solved.

Assessment: The right description of how marketplace businesses actually improve, and an implicit acknowledgment that no single 2026 launch reaccelerates growth. It also makes the investment case harder to underwrite: a strategy of hundreds of small improvements is unfalsifiable quarter to quarter and shows up only in the aggregate nights number. Which is why we anchor on nights growth and take rate rather than on the product-announcement cadence.

Guidance & Outlook

The Q4 revenue guide of $2.66–2.72B represents 7–10% year-over-year growth, a step down from Q3's 9.7% at the midpoint and roughly in line at the top end. Read alone that looks like deceleration. Management pre-empted the reading by pointing at the base period rather than the current one.

"Last year, we saw a meaningful acceleration of growth from Q3 to Q4. Our Q4 2025 guide takes this tougher comp into consideration. But on a year-over-2-year basis, we do anticipate a sequential acceleration from Q3 to Q4." — Ellie Mertz, CFO

Testing the comp defence: the claim is checkable and it holds. Q4 2024 nights grew 12% against Q3 2024's 8%, a four-point acceleration into the base period. Guiding mid-single-digit nights growth on top of a 12% comp implies a two-year stacked growth rate in the high teens, versus Q3 2025's 8.8% on top of 8%, or a stack in the high teens as well. The two-year construction is genuinely flat-to-better, not worse. We accept the framing.

Implied FY25 trajectory: nine-month revenue of $9,463M plus the $2,690M guide midpoint puts FY25 at approximately $12,153M, or +9.5% against FY24's $11,102M. Applying the approximately 35% adjusted EBITDA margin target gives roughly $4,254M of FY25 adjusted EBITDA. Since nine-month adjusted EBITDA is $3,511M, the FY target implies Q4 adjusted EBITDA of roughly $743M against $765M a year ago, a 2.9% decline. That is arithmetically consistent with the stated "flat to down slightly" and gives us confidence the margin guide is a real constraint rather than a rounded aspiration.

Guidance style: Airbnb has landed at or near the high end of its own revenue guide in each of the last several quarters, including this one, and the Q3 result was explicitly characterized as "the high end of our guidance range." We treat the Q4 range as conservatively set and model the upper half, with FY25 revenue at $12.15–12.20B.

What is missing: no 2026 revenue guide, no 2026 margin guide, no disclosed revenue or bookings figures for experiences, services or hotels, and no quantification of the incremental 2026 investment. Management committed to addressing 2026 on the February call. That call is now the single most important scheduled event in the story.

Analyst Q&A Highlights

How Much of the US Acceleration Is the New Payments Product, and What Does It Cost in Cancellations

The opening question went straight to the quarter's central attribution problem: management had credited Reserve Now, Pay Later with driving the US acceleration, and the questioner wanted the percentage and the cancellation offset. The answer supplied a striking attach rate but declined to size the contribution to nights growth, and conceded increased cancellations while asserting a positive net.

Q: "I guess you've mentioned the Reserve now Pay later quite a lot during the prepared remarks there. Just what percentage of the acceleration in the U.S. has come from that? Any early signs of what cancellation rates might look like on those bookings or your expectations of what they might look like?"
— Richard Clarke, Bernstein

A: "So about 70% of people that we offer Reserve Now, Pay later take us up on that offering. Now to your question of cancellations, we obviously tested this product pretty extensively before launching it in the U.S. to ensure that the benefit of the incremental lift in bookings from the payment offering was not more than offset by increased cancellations. So yes, there are increased cancellations, but we're highly confident that the net impact of the product is a lift to the net bookings."
— Ellie Mertz, CFO

Assessment: The question was answered sideways. A 70% attach rate among eligible guests tells you the feature is popular; it does not tell you how much of the two-point nights acceleration it produced, which is what an investor needs to know to judge whether the acceleration is repeatable. "Highly confident" about a net that was tested pre-launch is a reasonable answer from a company that will not disclose cohort economics, but it leaves the analyst unable to separate structural demand improvement from a one-time conversion step. We assume roughly half the US acceleration is attributable to the feature and therefore non-recurring, and we would revisit that if Q4 nights land at the high end of the mid-single-digit guide.

The Duration and Depth of the International Expansion Investment

A question probed how long the market-by-market expansion strategy takes to pay back and how much investment it absorbs, framed over a two-to-three-year horizon. The answer was unusually concrete about the sequencing, using Brazil as the mature case and Japan as the current one, and made clear that the timeline varies by market rather than following a single template.

Q: "As you look out over the next 2 to 3 years, can you talk a little bit about how investors should be thinking about the duration of execution and depth of investment needed to accomplish your goals with respect to international markets and improving sort of the density of the business beyond its core markets today?"
— Eric Sheridan, Goldman Sachs

A: "So I think our earliest expansion markets were in Latin America, Brazil being kind of the first key expansion market. And you've seen in that case, we have been at it for a handful of years, and we've been able to achieve not only significant market share in that market, but we continue to see incremental meaningful market share gains every year… A more recent target market is Japan. It was just a year ago that we began our efforts there in terms of, in particular, local marketing to make ourselves locally relevant, in particular for domestic stays."
— Ellie Mertz, CFO

Assessment: The most useful answer on the call for modelling purposes, because it converts a vague growth narrative into a dateable cohort structure. Brazil entered several years ago and still compounds share annually; Japan entered a year ago and is producing 27% domestic nights growth. That implies a roughly three-year lag from entry to visible national-level contribution, and it means the India cohort should become material around 2028. The investment is marketing-led, which is consistent with sales and marketing growing 24.3% against 9.7% revenue growth, and it also means the spend is discretionary and can be dialled back if 2026 margins come under pressure.

Whether Experiences and Services Contribute Yet, and What They Do to Retention

A two-part question asked whether the new offerings are contributing financially and whether bookers of those offerings show better retention or platform velocity. Management answered the funnel question with data and the retention question with an explicit "too early," while volunteering the three-to-five-year materiality timeline.

Q: "I'd love to hear an update on events and experiences. Both is it contributing any -- so far, do you expect a bigger contribution next year? And then second, people who do book those, are you seeing better retention or any signs that's helping the velocity of the overall platform?"
— Justin Post, Bank of America

A: "The first thing we're seeing is that a large percentage of people that are booking experiences, about half don't have an Airbnb stay associated with the reservation. And with services, for example, 10% of people have never even booked on Airbnb before… And as far as the retention velocity, it's a little bit early to see on retention. We really would want to see 12 months to really get a really good indicator of retention of what people use for Airbnb."
— Brian Chesky, Co-Founder & CEO

Assessment: Honest and correctly framed. The funnel data is the right leading indicator and it is positive: if half of experiences bookings are unattached and a tenth of services bookers are entirely new to the platform, the offerings are widening the top of funnel rather than redistributing existing demand. The retention answer is the right answer given a May launch, and it also means the key economic question, whether an experiences booker becomes a stays booker, is unanswerable until roughly mid-2026. Until then this line is spend without measurable return, which is precisely why the 2026 margin question matters so much.

Whether Hotels Are a Supply Backfill or a Standalone Business

A follow-up pressed the strategic ceiling of the hotels initiative, framing the sceptical case directly: if hotels exist only to fill gaps where home supply is thin, why would a guest choose Airbnb over an incumbent hotel-booking channel? The answer conceded that today it is largely backfill but argued the long-run answer is personalization rather than merchandising.

Q: "But do you see a path to really expand the hotel merchandising and booking experience beyond the backfill. We get the question a lot from investors like why would I go to Airbnb book a hotel, like is there a long-term case beyond the backfill, I guess, is the question?"
— Lloyd Walmsley, Mizuho

A: "Right now, it's a little bit more backfill filling and supply gap and prioritizing hotels when it's the last minute, say, night stay. But I think over time, it's really about personalized experience for every guest and understanding what kind of guests are you? What is your intent? And as we do that, I think we can expand our hotel offering a bit more."
— Brian Chesky, Co-Founder & CEO

Assessment: The candour is useful and the answer is weaker than the strategy deserves. Conceding "right now, it's a little bit more backfill" is accurate, but the stronger case was made elsewhere on the call and not here: Airbnb already owns the demand in the urban markets where its home supply is regulatorily capped, and monetizing that leaked traffic requires no incremental marketing. The personalization argument is a 2027 answer that depends on the AI roadmap landing first. For now we underwrite hotels as a supply-gap product with attractive incremental economics and no 2026 revenue contribution.

Whether the 2025 Investment Rolls Off in 2026

The most consequential exchange of the call, and the one where management gave the least. The question isolated the roughly $200M of 2025 investment in services and experiences and asked directly whether it persists into 2026, noting that the shareholder letter's margin language implied some of it might. The answer restated the letter's language, described 2025 as a heavy first-year launch, confirmed that experiences, services, hotels and AI will all be funded next year, and closed by deferring.

Q: "You guys are doing about, what, $200 million of incremental investments this year with some piece of that presumably rolling off next year. That said, your comments on margins for '26, while not specific, may suggest some of this $200 million proves sticky into next year. Can you maybe unpack that a bit more as to what sort of the incremental investment plan looks like for 2026?"
— Lee Horowitz, Deutsche Bank

A: "So you should anticipate that across Experience and Services, across hotels, across AI, we will be investing in those next year to drive growth. But where we sit today, we feel like we can afford those investments while continuing to deliver great core efficiencies from our core business and in doing so, deliver net-net a continued strength of margin. So more color on that to come."
— Ellie Mertz, CFO

Assessment: This is the exchange that sets our rating. The substantive disclosure is that 2026 carries four investment lines where 2025 carried two, offset by the absence of first-year launch costs. That is not obviously net-neutral, and the answer contained no number, no range, and no directional commitment beyond "continued strength of margin," a phrase compatible with anything from 33% to 36%. A management team that had 2026 margins comfortably in hand would have said so in November rather than February, particularly with the stock down 14.5% over twelve months. The deferral is the tell, and it is the reason we initiate at Hold rather than Outperform on an otherwise reasonable valuation.

Why Airbnb Sat Out the ChatGPT App-Integration Launch

A pointed question surfaced Airbnb's conspicuous absence from OpenAI's app-integration launch when other travel players participated, and asked why the company would forgo a new top-of-funnel surface while simultaneously building its own AI search. Management engaged fully rather than deflecting and articulated an explicit set of integration principles.

Q: "Airbnb was notably absent from ChatGPT's app integration launch when other major travel players were there. Can you just talk about your thought process here? And why you wouldn't explore that top of the funnel while also building out your own enhanced discovery and AI-powered search?"
— Douglas Anmuth, JPMorgan

A: "the simple answer is we just didn't think the integration was ready. We care a lot about how Airbnb shows up in the world… We didn't want to be positioned as essentially a data layer like a commodity… But that being said, I think with great integrations, you could absolutely see us on chatbots going forward."
— Brian Chesky, Co-Founder & CEO

Assessment: A clear articulation of a real strategic bet, and worth watching more closely than the market did. The company is wagering that its 90%-direct-and-unpaid traffic mix is durable enough that it can set terms on new discovery surfaces rather than accept them. That wager was correct for Google Hotel Finder, where declining preserved brand differentiation at trivial cost. It is a bigger wager against a general-purpose assistant that may become a primary discovery layer. The exchange did not resolve the question; it established that the absence was deliberate and that the door is open, which is the most an investor could reasonably expect this early.

Whether the Core Homes Business Can Reaccelerate to Low Double Digits

A question referenced management's September conference remarks suggesting the core business could reaccelerate, possibly into the low teens, and asked for elaboration. The answer restated the TAM argument, laid out an explicit supply-price-bookings flywheel, and made an unusually absolute claim about market maturity.

Q: "I wanted to just press a little bit more, Brian, on the idea of reaccelerating the business. I think you spoke at the SCIF Travel Conference in mid-September and talked about how you believe the business should reaccelerate. I think you even said into the low teens, although I don't want to misquote you. Maybe if you could just elaborate a little bit on that."
— Kenneth Gawrelski, Wells Fargo

A: "We're getting more homes in the top 300 markets around the world. As we get more supply, we'll, of course, sell more homes. But as we get more supply, prices will come down. As prices come down, we get more bookings… but we do not see a single market in the world that is truly mature. The only type of maturity is when there's a policy constraint, and there's just limits to how much supply you can add."
— Brian Chesky, Co-Founder & CEO

Assessment: The flywheel described is real and is the correct mechanism, but note what it implies: reacceleration is purchased with price. More supply lowers ADR, lower ADR lifts bookings, and GBV grows only if the volume elasticity exceeds one. Combined with the October migration to a single 15.5% host fee, which is also deliberately price-deflationary, management is explicitly steering toward a lower-ADR, higher-volume mix. That is almost certainly the right long-run trade against hotels. It also means investors should expect nights growth to outrun GBV growth and GBV growth to outrun revenue growth for several quarters, and should stop treating take rate as the health metric. Management did not repeat the low-teens number when given the chance, which we read as appropriate caution rather than retreat.

What They're NOT Saying

1. Any 2026 margin number, range, or floor: the question was asked directly and answered with "continued strength of margin" and a deferral to February. With four investment lines funded in 2026 against two in 2025, the absence of even a floor is the most conspicuous omission in the print.
2. Any revenue, GBV, or booking figure for experiences and services: six months post-launch, the company disclosed guest ratings, host applications, and funnel attachment, but not a single dollar. Management's own three-to-five-year materiality framing explains why, and simultaneously means investors are funding roughly $200M a year against an entirely undisclosed revenue base.
3. The size of the Reserve Now, Pay Later contribution to nights growth: asked explicitly for the percentage of US acceleration attributable to the feature and got an attach rate instead. Without it, the two-point nights acceleration cannot be decomposed into structural versus mechanical.
4. The cancellation rate on Reserve Now, Pay Later bookings: management confirmed cancellations are "increased" and the net is positive, but gave no rate. Nights are reported net of in-quarter cancellations only, and management flagged that these cancellations skew to dates nearer the stay, so some Q3 nights carry Q4 reversal risk of unquantified size.
5. Any hotel pilot metrics: three pilot markets, a custom-built product, and enthusiastic supply-side reception, with no room-night count, no commission rate, no take rate, and no revenue. "Best-in-class commission" was offered as a recruiting pitch, not as a disclosed rate.
6. What the October fee migration does to revenue: the shift of PMS hosts to a single 15.5% fee is presented purely as a host and guest benefit. Whether it is revenue-neutral, accretive, or dilutive was not addressed, and the Q4 guide of a roughly flat year-over-year take rate is the only indirect answer given.
7. The 2026 debt decision: $2.0B of long-term debt moved to current on the September balance sheet and neither the letter nor the call addressed whether it will be refinanced or repaid. At $11.7B of corporate cash this is not a solvency question, but it is a call on roughly a fifth of the cash balance that competes with the buyback.
8. Loyalty and advertising timelines: both were confirmed as active areas, with advertising explicitly gated on AI search landing first. No launch window was offered for either, and both have been "under consideration" across multiple prior calls.

Market Reaction

• Pre-print setup: ABNB closed at $120.53 on November 6, down 8.3% year-to-date against the S&P 500's +14.3%, down 14.5% over the trailing twelve months, and up just 0.6% over the trailing thirty days. The 52-week closing range entering the print was $105.69 to $161.42, putting the stock in the lower quarter of its own annual band.
• After-hours move: shares rose as much as 5% in extended trading on the revenue beat, the record adjusted EBITDA, and a Q4 guide the market read as solid.
• Next-day session (November 7): opened at $125.50, a 4.1% gap up, traded a $117.15 to $125.76 range, and closed at $120.88, up 0.3% or $0.35. The entire gap was surrendered intraday and the stock finished 3.7% below its own opening print.
• Volume: 11.8M shares against a 4.3M thirty-day average, or 2.7x normal, confirming the round trip was real repositioning rather than thin-tape drift.
• Benchmark: the S&P 500 closed the same session up 0.1%, so essentially all of the intraday move was idiosyncratic.

A stock that gaps 4% higher on a beat and closes flat on 2.7x volume has told you something specific: the buyers were positioning for the headline and the sellers were reading the detail. Three things in the detail plausibly did the work.

The margin trajectory outranked the volume acceleration. Adjusted EBITDA set a company record in absolute dollars and still compressed 238bp as a margin. Full-year margin guidance of approximately 35%, even raised from a 34.5% floor, sits roughly 140bp below FY24. A quarter that shows growth being purchased rather than earned is a quarter the market discounts, and the sequence of the disclosure mattered: the record EBITDA headline hit the wire first and the margin arithmetic emerged on the call.

The 2026 deferral landed badly. The exchange on whether the $200M of 2025 investment persists occurred during the Q&A, well after the after-hours pop had formed on the release. An investor who bought the headline at 4:15pm and heard "more color on that to come" at 5:30pm had a reason to reconsider by the next morning's open.

The positioning was not stretched into the print. Down 14.5% over twelve months and up 0.6% over thirty days, this was not a crowded long being unwound. That argues the fade reflects genuine reassessment of the 2026 setup rather than mechanical profit-taking, which is the less comfortable of the two explanations.

Street Perspective

Debate: Is the Nights Acceleration Structural or Mechanical?

Bull view: Nights accelerated two points to 8.8% with every major region steady or improving sequentially, which is not the signature of a single-product effect. Expansion markets are growing at double the core rate, app share of nights reached 62% with app nights up 17%, and 65 product improvements shipped in a single half-year. The acceleration is the cumulative return on three years of product and international investment finally compounding.

Bear view: The acceleration is concentrated in the US, management explicitly credited a payments feature launched in August, that feature attaches at 70% among eligible guests, and it carries an admitted increase in cancellations that Airbnb declined to size. Strip it out and the underlying rate is closer to Q2's. The company also guided Q4 nights to mid-single digits, which is a deceleration from the accelerating quarter it just reported.

Our take: The bear has the better of the near term and the bull the better of the medium term. Reserve Now, Pay Later is a real conversion improvement and a real one-time step; we assume roughly half the US acceleration is attributable to it and therefore non-recurring, which laps in Q3 2026. The regional breadth and the expansion-market cohort data are genuine and independent of it. We model nights growth of 6–8% through 2026 rather than the low double digits the reacceleration narrative implies.

Debate: Are the Three New Businesses an Option or a Tax?

Bull view: Airbnb is converting a single-product marketplace into a travel platform while the core still generates $4.5B of trailing free cash flow. Experiences already brings genuinely new users (half of bookings unattached to a stay, a tenth of services bookers new to the platform entirely), hotels monetizes traffic the company already owns at near-zero incremental demand cost, and AI has already produced measurable operating leverage. Each is a plausible multi-billion-dollar line, and the market is paying nothing for any of them at 15x EV to free cash flow.

Bear view: Three simultaneous launches cost roughly $200M in 2025 with more committed in 2026 and produce zero disclosed revenue. The CEO put experiences three to five years from materiality; hotels is explicitly "a little bit more backfill" today; AI search has not shipped. Meanwhile the core is growing high single digits with a compressing take rate, and every dollar of new-business investment is a dollar not returned through a buyback the company is executing at an attractive multiple.

Our take: Both are right about different lines and the market is wrong to price them identically. Hotels has the cleanest economics because it inverts the cold-start problem and requires no demand spend. AI customer support is already earning its keep in the operations and support line. Experiences is the expensive one, with the longest payback and the softest evidence, and it is where the $200M mostly went. We would credit hotels and AI in the base case and treat experiences as an unpriced option rather than a modelled contributor. The aggregate is closer to an option than a tax, but only just, and the balance tips on what February's margin guide reveals.

Debate: Is 15x EV to Free Cash Flow Cheap, or a Value Trap on a Decelerating Asset?

Bull view: At $120.88 the shares carry roughly $68.4B of enterprise value against $4.55B of trailing free cash flow, a 15.0x multiple and a 5.8% free cash flow yield on the equity, for a business with 50% peak-quarter EBITDA margins, $9.7B of net cash, no capital intensity, and a buyback consuming three quarters of free cash flow. The tax rate is about to fall from roughly 21% to the mid-to-high teens for structural reasons. The stock is down 14.5% over twelve months against a market up 14.3% year-to-date, and the de-rating has already discounted the deceleration.

Bear view: The multiple is optically low because free cash flow is flattered by float. Airbnb collects service fees at booking and recognizes revenue at stay, so working capital grows with GBV; trailing free cash flow margin of 38% against a net income margin near 22% is the gap, and it compresses if GBV growth slows. Behind the float, this is a high-single-digit revenue grower with a declining take rate, compressing EBITDA margin, no 2026 guide, and three cash-consuming projects. Fifteen times free cash flow is the right multiple for that, not a discount to it.

Our take: The float point is correct and materially under-appreciated, but it cuts less than the bear implies. Unearned fees of $1,820M and funds held on behalf of customers of $7,209M grow with GBV and would stop contributing if GBV growth went to zero. GBV is growing 14%, so the float is a tailwind for the foreseeable modelling horizon rather than a reversing one. The more honest bear point is simpler: at 15x free cash flow on a business compounding high single digits, you are paid roughly a market return for accepting single-name risk plus an unresolved margin question. That is the definition of a Hold. The multiple protects the downside; it does not create the upside.

Model Update & Valuation Framework

This is our initiation on ABNB, so there is no prior model to revise. The table below sets our starting estimates and the driver assumptions behind them.

Valuation framework: At the $120.88 close and a fully diluted count of 646M, the equity is worth approximately $78.1B. Corporate cash and investments of $11,719M less $1,998M of debt gives $9,721M of net cash, for an enterprise value of roughly $68.4B. Against trailing twelve-month revenue of $11,943M that is 5.7x sales; against trailing free cash flow of $4,550M it is 15.0x; against our FY25E adjusted EBITDA of roughly $4.25B it is 16.1x. The free cash flow yield on the equity is 5.8%.

Why the base case is 16x: the multiple has to sit between what a high-single-digit compounder with an unresolved margin question deserves and what a 50%-peak-margin, capital-light, net-cash marketplace with a real buyback deserves. Sixteen times forward free cash flow is roughly where the shares have traded through the deceleration and does not assume any re-rating. On $4.85B of FY26E free cash flow that gives $77.6B of enterprise value; adding net cash and dividing by 646M diluted shares gives $135. From $120.88 that implies roughly 12% upside.

Risk-reward: base $135 (+12%), bull $167 (+38%), bear $100 (−17%). The bull-to-bear span is roughly 2.2:1, which sounds attractive until you note that the base case alone is a market-like return and that the bull case requires two things to go right simultaneously (volume reacceleration and margin defence) that management has explicitly declined to underwrite for 2026. The asymmetry is not sufficient for Outperform at this price. It would be at roughly $105, near the bottom of the 52-week closing range, where the base case implies close to 30%.

Thesis Scorecard Post-Earnings

This is our initiation, so the table below establishes the thesis rather than grading it. These are the pillars we will score every quarter from here.

Overall: Thesis established. The bull case rests on cash generation and a volume acceleration that is real but partly purchased; the bear case rests on a compressing monetization wedge and an investment cycle that widens in 2026 before any adjacency contributes revenue. Both sides are evidenced in this quarter, which is why the shares gapped up and closed flat.

Action: Hold. The valuation is supportive rather than compelling at 15.0x trailing free cash flow with $9.7B of net cash, and the downside is reasonably protected by the buyback and the balance sheet. But the base case is a market-like return and the one event that could resolve the margin debate is the February call. We would upgrade on a 2026 adjusted EBITDA margin guide at or above 35% with nights holding mid-to-high single digits, or on a move toward $105 without a change in fundamentals. We would downgrade on a 2026 margin guide below 34%, on nights decelerating through mid-single digits, or on evidence that the October fee migration is revenue-dilutive rather than neutral.