The Margin Answer Arrives Intact: FY26 Guided to Accelerate on Stable Margins, With Three Pricing Changes Doing 300 Basis Points of the Work

Results vs. Consensus

Q4 2025 Scorecard

Year-Over-Year Comparisons — Q4

Full-Year 2025 vs. 2024

Q-sum check: the four reported 2025 quarters sum exactly to the full-year figures. Revenue $2,272M + $3,096M + $4,095M + $2,778M = $12,241M. Net income $154M + $642M + $1,374M + $341M = $2,511M. Adjusted EBITDA $417M + $1,043M + $2,051M + $786M = $4,297M. Free cash flow $1,781M + $962M + $1,349M + $521M = $4,613M. Nights 143.1M + 134.4M + 133.6M + 121.9M = 533.0M. No reconciling items.

Quality of the Print

Revenue: The 12.0% growth rate came against the year's hardest comparison, with Q4 2024 having accelerated to 12% nights growth from Q3 2024's 8%. Ex-FX growth was 11%, so currency supplied roughly a point. The genuinely new information is the composition disclosed on the call: management estimated that three product changes shipped in and around the quarter contributed more than 200 basis points of nights growth and roughly 300 basis points of GBV growth. That is the first time this company has quantified a product contribution to a growth rate, and it should be read as a confidence signal rather than a hedge, because the natural incentive is to leave the acceleration attributed to demand.

Margins: The operating margin decline of 766bp is arithmetically real and analytically hollow. Working the bridge from the adjusted EBITDA reconciliation: operating income fell $161M while total add-backs grew $182M and adjusted EBITDA grew $21M, and 182 minus 21 is 161. The single largest component is the lodging, withholding and transactional tax line, which was an $81M add-back this quarter against a $43M credit a year ago, a $124M swing that management sized at roughly $90M of one-time non-income tax matters. Stock-based compensation added another $43M of the gap. Underneath the add-backs, adjusted EBITDA margin of 28.3% against 30.8% is the honest number, and 256bp of compression on a quarter carrying deliberate growth investment is a tolerable outcome.

EPS: Three below-the-line items and one offset again. Interest income fell 11.5% to $162M, other income swung to a $38M expense from $9M of income, and the two together cost roughly $68M of pre-tax income against a $393M base. Running the other way, the effective tax rate collapsed to 13.2% from 25.9%, saving $109M of provision on a smaller pre-tax base, which is the One Big Beautiful Bill Act benefit arriving a quarter earlier than the 2026 start date management had signposted. The buyback removed 22M shares from the diluted weighted average. Net of all of it, GAAP EPS of $0.56 is a poor representation of a quarter in which adjusted EBITDA grew and free cash flow grew 13.8%.

Key Business Metrics

Regional Performance

EMEA — the quarter's genuine regional surprise

EMEA accelerated from mid-single-digit nights growth in Q3 to high-single-digit in Q4, and management attributed it to broad demand improvement across both core and expansion markets, naming the UK, Spain and Italy. That matters more than the Latin America deceleration because EMEA is a far larger share of the business and had been the least dynamic region for several quarters. The 12% reported ADR gain is mostly currency, with only 4% surviving ex-FX, so the volume acceleration is the signal here rather than the value.

Assessment: The most credible evidence in the release that demand improved independent of the product changes, since none of the three quantified features were EMEA-specific. Reserve Now, Pay Later was US domestic only in Q4; the cancellation-policy change was global but modest; the fee migration touched property-manager listings. An EMEA acceleration of that shape is a macro and awareness signal, and it supports management's characterization of a slightly better macroeconomic environment than anticipated.

Latin America and Asia Pacific — still the growth engine, decelerating in one case

Latin America decelerated from low-20s nights growth to high teens, the only region to slow. Brazil remains the anchor with origin nights up more than 20% and first-time bookers up 17%, and the CEO disclosed that Brazil has moved from a top-ten to a top-five market and was the second-largest contributor to first-time bookers in Q4 behind only the United States. Mexico accelerated to high-teens origin nights growth. Asia Pacific held at mid-teens, with India origin nights up 50% and first-time bookers up more than 60%.

Assessment: The Latin America deceleration is the natural consequence of Brazil maturing into a top-five market; a market that large cannot compound at low-20s indefinitely, and the Mexico acceleration is the intended succession. The India figures are the most striking in the release and are earlier-stage than they look, since first-time booker growth of 60% is a leading indicator that precedes revenue by roughly two to three years on the Brazil template. Asia Pacific remains the region with the largest gap between penetration and opportunity.

North America — stabilized, not reaccelerated

North America held mid-single-digit nights growth for a second consecutive quarter. Management disclosed that growth in the region was low single digits in Q1 and Q2 of 2025 before accelerating in Q3 and again in Q4, and characterized the region as one of the underpinnings of its 2026 optimism. This is also the region where Reserve Now, Pay Later was live, so the mid-single-digit print embeds most of the quantified product lift.

Assessment: The trajectory is right but the level is unremarkable, and the product-lift disclosure means the underlying North American rate is closer to low-to-mid single digits. This is the region where the FY26 acceleration guide is most exposed: it is the largest market, it carries the highest ADR, and its improvement to date is the most feature-dependent. The global rollout of Reserve Now, Pay Later to new markets and to US cross-border stays is management's answer, and it is a credible one.

Key Topics & Management Commentary

Overall Management Tone: Confident and, for the first time in our coverage, quantitative where it counts. The prepared remarks were structured as an argument rather than a recitation, walking from platform rebuild to innovation model to results, and the CFO put explicit basis-point figures on the product contribution to growth and on the cancellation cost of the payments feature. The contrast with November, when the same team answered a direct question on 2026 margins with a directional phrase and a deferral, is the most important tonal fact of the call. Where management remained vague was on the composition of the second-half 2026 growth levers that replace the ones now lapping.

1. The FY26 Guide: Accelerating Revenue on Stable Margins

Management guided full-year 2026 revenue growth to accelerate to at least low double digits against FY25's 10.3%, and adjusted EBITDA margin to be stable year-over-year against FY25's 35.1%. In November the same question produced a directional assurance about continued margin strength and an explicit deferral to this call. Here it produced a commitment, delivered in the prepared remarks rather than extracted in Q&A.

"We expect revenue growth to accelerate to at least low double digits in 2026. We expect adjusted EBITDA margins to be stable year-over-year. And we'll do all of this without investing billions or tens of billions of dollars." — Brian Chesky, Co-Founder & CEO

The CFO framed the same commitment from the cost side, describing cost of revenue and operations and support as scaling roughly linearly with revenue, with the incremental investment concentrated in sales and marketing (both programmatic and go-to-market supply acquisition) and product development.

"our ambition is to accelerate the top line, and we're giving ourselves the flexibility within the stable margins compared to last year to invest to achieve that acceleration." — Ellie Mertz, CFO

Assessment: This is the disclosure that changes the rating. Accelerating revenue on flat margins is a materially different proposition from the FY25 pattern of decelerating margins funding growth investment, and it removes the specific uncertainty we cited in November as the reason to wait. The commitment is not costless to management: a stable-margin guide constrains the pace at which experiences, hotels and AI can be funded, which means the company has chosen a disciplined path over an expansive one. We take the guide at face value and note the check on it is Q2 2026, when the H1 product lift begins to lap.

2. Three Product Changes Did 300 Basis Points of the GBV Growth

The CFO attributed the Q4 acceleration to three specific changes: the launch of Reserve Now, Pay Later, the October revision of cancellation policies, and the beginning of the migration to a simplified single service fee. She then sized them, estimating that the three together delivered more than 200 basis points of nights growth and roughly 300 basis points of GBV growth in the quarter. Against reported growth of 9.8% and 15.9% respectively, that leaves underlying rates near 8% and 13%.

The mechanism on the fee migration is the most interesting of the three, because it worked through price rather than through conversion friction.

"in making the migration, what we found is that many of the hosts did not take up their rate. Instead, the effective ADR to guests came down modestly, which obviously you can conclude is really great from an affordability perspective as well as elasticity and that is the reason that has been a driver or contributor to growth in Q4." — Ellie Mertz, CFO

Assessment: Disclosing this is the right call and an unusually honest one, since the alternative was to leave a 16% GBV quarter attributed to demand. It also reframes the investment case in a useful way. If three modest, cheap, entirely controllable product changes can add 300 basis points to GBV growth in one quarter, then the growth algorithm is more elastic to product than the last three years of high-single-digit results implied, and management's "hundreds of improvements" framing has a measurable payoff. The offsetting risk is arithmetic: these three lap in H2 2026, and the FY26 guide requires replacements. Management named the global rollout of Reserve Now, Pay Later, expansion of the fee migration to individual hosts, incremental supply, and expansion markets. That list is plausible but unquantified.

3. The Reserve Now, Pay Later Cancellation Cost, Finally Sized

In November management confirmed that the payments feature raised cancellations and declined to quantify it, which we flagged as the largest unanswered question in that print. This quarter the number arrived, and it is small.

"in terms of the aggregate nominal increase in cancellations rate, it's approximately 1%. So an average of maybe 16% cancellation rate historically going to 17%. It's obviously higher within the cohort that chooses that product, but it's not hugely material relative to the broader cancellations on the platform." — Ellie Mertz, CFO

Two secondary effects were disclosed alongside it. The feature lengthens booking lead times, which management frames as competitively useful, and it carries what the CFO described as a modestly positive impact on ADR, because guests who do not have to put a large purchase on a credit card at the point of booking trade up to more expensive listings.

Assessment: A one-point rise in a 16% base rate against a 200-plus basis point lift in nights is a favourable trade by any reading, and the disclosure of the platform-level cancellation rate itself is new and useful. The ADR mechanism is the part we had not modelled: removing the upfront cash constraint pushes guests toward larger homes, which is why Q4 ADR rose 5.9% with North America's gain attributed to mix shift toward four-bedroom-plus entire homes. That makes the feature a GBV lever twice over, on volume and on price, which explains why it contributed more to GBV growth (roughly 300bp with the other two changes) than to nights growth (more than 200bp).

4. Take Rate: Down Again in Q4, Guided Up in Q1

Implied take rate was 13.6% against 14.1% a year earlier, a second consecutive year-over-year decline, attributed as in Q3 to currency and to the timing gap between booking and stay. The forward guide is the change: management expects the Q1 2026 implied take rate to be up slightly year-over-year.

"in terms of the implied take rate, it should be modestly above where we were in Q1 of last year, mostly due to some timing consideration." — Ellie Mertz, CFO

The Q1 arithmetic is internally consistent and worth checking. Revenue is guided to grow 14–16% while GBV is guided to grow in the low teens, which requires roughly one to three points of take-rate expansion. Management identified the components: a currency tailwind of nearly three percentage points, revenue recognition in Q1 from the longer-lead-time bookings made in Q4, and an Easter timing effect worth about 50 basis points of incremental Q1 revenue and 50 basis points less in Q2.

Assessment: The guided expansion is largely mechanical rather than structural, which is the correct way to read it and still an improvement. Two of the three drivers (currency and Easter) reverse; the lead-time effect is a level shift that persists while lead times keep lengthening. What matters for the thesis is that the fee migration, which we flagged in November as a deliberate monetization concession, is not showing up as take-rate destruction. That was Bear 1's most concrete near-term risk and it has not materialized.

5. Project Hawaii and the Innovation Model

The CEO spent most of his prepared remarks on the mechanism behind the acceleration rather than the acceleration itself, describing a small dedicated team created two years ago with a mandate to make it easier to find and book a home, starting with search filters and booking-flow tweaks and escalating to a full checkout redesign.

"Now these are just a few of the hundreds of improvements the team shipped, driving hundreds of millions of dollars in revenue in 2025 alone. And we believe Project Hawaii will deliver hundreds of millions more this year." — Brian Chesky, Co-Founder & CEO

The model was then described as having been extended to pricing, supply, international expansion and new businesses, with the pricing workstream expected to contribute comparably in 2026.

Assessment: "Hundreds of millions of dollars in revenue" against a $12.2B base is roughly two to four points of growth, which is consistent with the 200 to 300 basis point product attribution disclosed by the CFO and gives the two figures mutual corroboration. The framing is also a partial answer to the November criticism that a strategy of hundreds of small improvements is unfalsifiable: management is now attaching revenue to the programme and naming the next one. We would still prefer a disclosed run-rate to a range, but this is a real improvement in accountability.

6. AI: Roughly a Third of North American Tickets, and No Capex

The customer-support deployment progressed materially. The shareholder letter reports that AI-powered support is now live for English, French and Spanish speakers across the United States, Canada and Mexico, and that about a third of issues are resolved without needing an agent when users message the assistant. The CEO gave a slightly different cut of the same metric on the call, putting nearly 30% of English-language North American tickets as handled by an AI agent, with global rollout planned this year and voice support intended to follow chat.

On the question that most concerns investors funding an AI transition, the answer was unusually blunt.

"one of the great things about Airbnb is that we have a very, very cost-efficient innovation model. So unlike other companies, we're not building models. We do not have a huge CapEx cost base. So our investment in AI will not affect the P&L. I don't think you'll see it in the P&L." — Brian Chesky, Co-Founder & CEO

Management also hired a chief technology officer to lead the effort, Ahmad Al-Dahle, previously sixteen years at Apple and most recently leading the generative AI team at Meta that built the Llama models. More than 80% of Airbnb engineers now use AI tools.

Assessment: The strategic asymmetry here is real and under-priced. Airbnb is capturing the cost side of AI (support deflection, engineering productivity) without the capital side, in a market that is currently penalizing companies for AI capital intensity. The Q3 note credited the support deflection when operations and support fell in absolute dollars; this quarter that line grew 12.8%, slightly faster than revenue, so the leverage did not repeat in Q4. We read that as the cost of geographic and language expansion of the assistant rather than a reversal, and expect the leverage to resume as the deployment matures. The CTO hire is a substantive signal, not a press release: this is a senior operator from the team that shipped a frontier model family.

7. The Disintermediation Question, Answered More Confidently Than in November

Three months ago management explained why it declined to join OpenAI's app-integration launch, with a defensive framing about not becoming a data layer. This quarter the same subject was handled as an offensive argument, built on assets rather than on principles: 200 million verified identities, 500 million proprietary reviews, messaging used by 90% of guests, global payment processing, customer support and insurance, and more than $100 billion of payments handled through the platform.

"what we've seen is I think they're going to be positive for Airbnb. And I'm very, very deep in this space. And what we see is that traffic that comes from chatbots converts at a higher rate than traffic that comes from Google." — Brian Chesky, Co-Founder & CEO

The underlying thesis was restated in the same terms used in November, that model access is not a moat because every frontier model is available to every company, so specialization is where travel is won.

Assessment: The conversion datapoint is the substantive addition and it is a meaningful one, because it inverts the bear framing. If chatbot-referred traffic converts better than Google-referred traffic, assistant-mediated discovery is a channel-mix improvement rather than a disintermediation threat, at least at current volumes. The caveat is that it is a single unaudited datapoint at what must still be very small absolute volume, and that the economics of that channel are unset. We treat the risk as reduced but not retired, and we would want to see the conversion claim repeated with a volume figure attached.

8. Hotels: Sized for the First Time

The hotels pilot expanded from three cities to four with the addition of San Francisco, and management sized the business for the first time: as of Q4 hotels were a single-digit percentage of total nights booked, growing at nearly twice the rate of the overall platform, with plans to expand supply through the year and exit 2026 at a meaningfully larger share. In New York alone the platform now carries more than 100 hotels and more than 20,000 rooms.

The strategic framing also widened materially from November's supply-gap logic.

"What we've now evolved to is a much bigger strategy, a much more expansive strategy. It turns out, obviously, as we spend a lot of time with our guests that a lot of guests love to book homes and hotels." — Brian Chesky, Co-Founder & CEO

Management stayed deliberately with boutique and independent properties rather than chains, on the grounds that independents pay higher commissions than chains and have been actively seeking an additional distribution channel.

Assessment: Upgrading our November read. A single-digit percentage of nights growing at roughly twice the platform rate is already a real contributor rather than an experiment, and the commission argument is the part investors should note: independents pay more than chains, so the mix Airbnb has chosen is the higher-take-rate half of the hotel market, not the residual half. The unresolved item is the connectivity plumbing, which was raised on the call and answered with strategy rather than with a technical commitment. Property-management-system integration and multi-rate support are genuine engineering obligations, and they are the practical constraint on how fast supply can scale.

9. International Expansion: Brazil Graduates, India Arrives

The CEO put a sharper number on the concentration than the letter does, disclosing that roughly 70% of revenue comes from just five countries. Brazil is now the template: a focused team, local payment methods including interest-free installments, cultural campaigns around Carnival, and a move from a top-ten to a top-five market. In Q4 Brazil was the second-largest contributor to first-time bookers behind only the United States.

India is the newest cohort, with origin nights up 50% and first-time bookers up more than 60%, and management explicitly signalled an intent to accelerate it further in 2026.

Assessment: The clearest illustration yet of the cohort structure we described in November: Brazil entered several years ago and has now graduated into a top-five market that also seeds first-time bookers globally; Japan entered in Q4 2024 and is producing domestic acceleration; India entered most recently and is producing the largest percentage figures on the smallest base. With 70% of revenue in five countries, the arithmetic ceiling on this strategy is high enough that it does not bind within any reasonable forecast horizon. This remains the most under-discussed pillar of the story.

10. Events as a Supply Acquisition Channel

Management reframed large events as primarily a supply-recruitment mechanism rather than a demand event, using the 2024 Paris Olympics as the case study: more than 40,000 listings were added for the Games, and the CEO described the same playbook as now being repeated for the World Cup. The Milan Winter Olympics are live now and the FIFA World Cup arrives this summer across 16 North American cities.

The CFO was careful to bound the direct financial contribution.

"I would say those two events, in particular, they are large events on the platform, but in scale are very small portions of the overall business." — Ellie Mertz, CFO

Assessment: The supply-acquisition framing is the correct one and the CFO's caveat is the right guard against modelling a World Cup revenue bump. The durable value is threefold: hosts recruited for an event who keep hosting, a policy reset with municipal governments that reframes Airbnb as capacity rather than nuisance, and brand awareness. The first is measurable in principle, but no retention data on the Paris cohort was disclosed, so the durability of event-recruited supply remains an assertion rather than a demonstrated result. We assume no material in-period revenue contribution from either event and treat any as upside.

11. Capital Allocation: More Than 80% of Free Cash Flow Returned

Airbnb repurchased $1.1B of Class A stock in Q4, up from $857M in Q3, and $3,789M across 2025.

"in 2025, we repurchased $3.8 billion of our common stock using over 80% of our free cash flow. Returning capital to shareholders remains a key component of our capital allocation strategy. Since introducing our share repurchase program in 2022, we've reduced our fully diluted share count by about 9%." — Ellie Mertz, CFO

The remaining authorization is $5.6B. The company ended 2025 with $11.0B of corporate cash and investments, $7.0B of funds held on behalf of guests, and $1,999M of debt now in current liabilities. Cumulative free cash flow since the IPO is approaching $19B.

Assessment: The payout ratio rose from roughly three quarters of free cash flow at Q3 to more than 80% for the full year, executed into a share price that fell 17.2% in the thirty days before this print. That is the right direction on the right tape. The caveat from November stands and is slightly worse: stock-based compensation of $1,592M in 2025 grew 13.1%, faster than revenue, so a meaningful share of the $3.8B is offsetting dilution rather than shrinking the base. Management did commit that 2026 growth in stock-based compensation and headcount will be lower than 2025, which is the right acknowledgment. The $2.0B of current debt remains an unaddressed 2026 call on cash.

Guidance & Outlook

Testing the Q1 guide for internal consistency: revenue is guided to grow 14–16% while GBV grows in the low teens, which mathematically requires take-rate expansion of roughly one to three points, and management guided exactly that. Three components were named: a currency tailwind of nearly three percentage points, revenue recognized in Q1 from the longer-lead-time bookings that Reserve Now, Pay Later pulled into Q4, and Easter falling on April 5, which management sized at about 50 basis points of incremental Q1 revenue and 50 basis points less in Q2. The guide holds together. It is also worth noting that on this construction the Q1 beat versus the Street's $2.53B is partly a calendar and currency artifact rather than a pure demand upgrade.

The FY26 acceleration, decomposed: FY25 grew 10.3%. To reach "at least low double digits" with FX fading through the year, as management explicitly flagged, requires the underlying rate to improve rather than merely be translated upward. The levers named were the continuation and global expansion of the Q4 product changes, expansion of the fee migration from property managers to individual hosts, incremental supply acquisition, and expansion markets. The lever conspicuously not counted is AI search, which management stated is currently in pilot with nothing baked into the outlook.

The lapping problem, stated plainly: the three features that delivered more than 200 basis points of nights growth and roughly 300 basis points of GBV growth in Q4 will annualize in H2 2026. The CFO acknowledged this directly, saying the company will be comping those launches in the back half and intends to have incremental growth levers to support the guide. That is the honest version of the risk and the single most important thing to monitor across the next two prints. A Q3 2026 nights growth print that falls back toward the underlying 8% would put the FY26 guide at risk.

What the margin guide constrains: holding adjusted EBITDA margin flat at approximately 35% on revenue of roughly $13.6–13.8B means adjusted EBITDA of approximately $4.75–4.85B, or about $460–560M of incremental EBITDA on about $1.4–1.6B of incremental revenue. The CFO described cost of revenue and operations and support as scaling roughly linearly, with the incremental spend going to sales and marketing and product development. That is a coherent plan and it is also a real constraint: it caps how quickly hotels, experiences and the AI rollout can be funded, which is exactly the discipline that was missing from the November framing.

Analyst Q&A Highlights

Whether AI Platforms Could Build a Competing Rental Marketplace, and Whether Economics Get Shared

The opening question was the disintermediation question in its sharpest form: given the pace of AI development, why could a general-purpose assistant not simply launch its own short-term-rental platform, and is there a risk Airbnb ends up sharing economics with a discovery layer it does not control? The answer built an asset-based rather than principle-based defence and volunteered a conversion datapoint that had not previously been disclosed.

Q: "given the sort of speed of innovations going on, why do you think those AI platforms couldn't launch a short-term rental platform over time? And maybe secondly, do you see any risk that you'll have to share your economics with an AI platform at some point going forward? Or do you expect you'll be able to retain the same level of direct traffic you have today in an AI world?"
— Richard Clarke, Bernstein

A: "I think these chatbot platforms are going to be very similar to search. They're going to be really good top-of-funnel discoveries. And in fact, what we've seen is I think they're going to be positive for Airbnb… what we see is that traffic that comes from chatbots converts at a higher rate than traffic that comes from Google. But the other thing to know, and this is the most important point, is that these models are not proprietary."
— Brian Chesky, Co-Founder & CEO

Assessment: A stronger answer than the equivalent exchange in November, and the conversion claim is the reason. Framing assistant traffic as a better-converting version of Google traffic reframes the entire debate from defence to channel mix. The claim is unaudited and the absolute volume is presumably tiny, so it does not settle the question. What it does establish is that management is now measuring the channel rather than declining to participate in it, which is a different posture from the one that skipped the app-integration launch.

How Reserve Now, Pay Later Cancellations Are Pacing, and How They Are Baked Into the FY26 Guide

The most valuable exchange of the call for modelling purposes. The question asked for the cancellation pacing against expectations, including under Q1 weather disruption, and how cancellation assumptions enter the full-year EBITDA guide. The answer produced the platform-level cancellation rate, a figure this company had not previously disclosed.

Q: "can you give us a sense of how Reserve now, pay later cancellations have been pacing relative to your expectations, perhaps particularly in the face of weather disruptions in the 1Q? And then how you're thinking about baking in cancellation expectations to your full year adjusted EBITDA guide?"
— Lee Horowitz, Deutsche Bank

A: "in the segments that we have launched this offering, the cancellation curves have been very close to what we saw from a tested perspective… I think one piece of perspective is that in terms of the aggregate nominal increase in cancellations rate, it's approximately 1%. So an average of maybe 16% cancellation rate historically going to 17%."
— Ellie Mertz, CFO

Assessment: This retires the largest open item from the November note. A one-point increase on a 16% base, in exchange for more than 200 basis points of nights growth, is a clearly favourable trade, and the confirmation that realized cancellation curves match the pre-launch test set removes the tail risk that the feature behaves worse at scale than in testing. The disclosure of the base rate itself is a bonus, because a 16% platform cancellation rate is now a durable modelling input for cancellation-sensitivity work that was previously guesswork.

Whether the AI Build Shows Up in the P&L, and What Success Looks Like in a Year

A two-part question asked what areas of AI improvement would be visible twelve months out and what the gross-margin impact of increased AI investment would be relative to last year. Management answered the cost question first and categorically, then laid out four specific twelve-month markers.

Q: "if we're sitting here a year from now, what are the areas you're most focused on or seeing improvements to the platform using AI this year? That's one. And then two, maybe one just on the P&L impact. Any help at all on how you're thinking about the impact on gross margins from increased AI investment this year versus last year?"
— Brian Nowak, Morgan Stanley

A: "unlike other companies, we're not building models. We do not have a huge CapEx cost base. So our investment in AI will not affect the P&L… Right now, nearly 30% of tickets in North America that are English-based are handled by an AI agent. A year from now, if we're successful, significantly more than 30% of tickets will be handled by a customer service agent in many more languages… AI customer service will not only be chat, it will be voice."
— Brian Chesky, Co-Founder & CEO

Assessment: The most investable AI answer we have heard from a consumer-internet management team this cycle, because it is the only one that describes AI as a margin input rather than a capital commitment. The four markers given (higher ticket deflection, more languages, voice as well as chat, and engineering productivity) are all cost-side and all measurable. Note the tension with Q4's actuals: operations and support grew 12.8%, faster than revenue, so the deflection has not yet produced net leverage at the consolidated line. We read that as the cost of scaling the assistant across geographies and languages, and we expect the leverage to show in H2 2026 as the rollout matures. If it does not, this answer becomes the thing to press on.

Where the Incremental 2026 Spend Goes If Margins Stay Flat

The sharpest financial question of the call, and one that did the arithmetic in public: holding margins flat on the guided revenue acceleration implies roughly $800M of incremental year-over-year spend, so where does it go? The answer walked the P&L line by line and, unusually, named supply acquisition for the new businesses as a distinct bucket.

Q: "even at the low end of your revenue guidance to keep margins flat, you have to figure out a way to spend some $800 million more year-over-year. So just wondering where the larger spend buckets are going to be for this year."
— Stephen Ju, UBS

A: "where you will see some incremental investment to drive growth is obviously in sales and marketing. This is both in the form of programmatic marketing, but more so in terms of our go-to-market efforts. What this means is all of our efforts around acquiring supply, not just homes, but obviously also for experience services and hotels. And then we will also continue to grow our investments in product development to allow for a greater accelerated pace of innovation."
— Ellie Mertz, CFO

Assessment: The answer is specific enough to model and it changes where we expect the pressure to show. Sales and marketing already grew 27.1% in Q4 against 12.0% revenue growth, and the incremental dollars are going into the same line, weighted toward supply acquisition rather than demand generation. That is the better of the two uses, since supply acquisition compounds into a durable asset while performance marketing rents demand. The framing of margins as flexibility to be spent rather than a floor to be defended is the correct posture for a company trying to reaccelerate, and it is also the reason we would treat a margin guide cut in 2026 as a genuine thesis event rather than a rounding issue.

When Hotels Move the Needle, and What Is Boosting the Q1 Take Rate

A two-part question asked for a realistic timeline on hotels becoming material to revenue growth, then pressed on an apparent inconsistency in the Q1 guide: revenue growth accelerating versus Q4 while nights growth and ex-FX bookings growth both look flat to slightly slower. The answer sized hotels for the first time and then itemized the Q1 take-rate mechanics.

Q: "your revenue growth is sort of accelerating versus Q4, but your room night growth seems like it's slightly decelerating or similar growth and your bookings growth ex FX is slightly decelerating. So is there something that's boosting take rate in Q1 that caused that revenue growth to accelerate?"
— Mark Mahaney, Evercore ISI

A: "as of Q4, hotels was a single-digit percent of total nights booked but growing nearly double that of the overall platform… As we called out in the letter, the realized tailwind of FX in Q1 will be quite strong at nearly 3 percentage points. We're also getting the benefit of earlier lead times of bookings in Q4 that will realize in stays and hit revenue in Q1… But we anticipate it will support about 50 basis points of incremental revenue in Q1 and 50 basis points less revenue in Q2."
— Ellie Mertz, CFO

Assessment: The question was right to be suspicious and the answer is honest. The Q1 revenue acceleration is substantially composition rather than demand: currency, a booking-to-stay timing benefit created by the lead-time lengthening that Reserve Now, Pay Later caused, and Easter falling in Q1. Two of those three reverse. Investors modelling Q2 2026 should carry the 50 basis point Easter reversal and should not extrapolate the Q1 take-rate expansion. Separately, sizing hotels at a single-digit percentage of nights growing at nearly twice the platform rate was the release's most useful unprompted disclosure, and it is well ahead of where we had the business in November.

What the Single-Fee Migration Actually Did, and Whether All Hosts Move Over

A question probed the results of the October migration of property-management-system-connected hosts to an all-in commission, and whether the structure would eventually extend to the whole host base. The answer explained the problem being solved (cross-listing price inconsistency), the observed guest-side outcome, and confirmed a pilot for individual hosts.

Q: "I wanted to ask about the new all-in commission structure for PMS connected host. What are some of the benefits you're seeing from that change? And could you see Airbnb moving all of its hosts over to that structure longer term?"
— Kevin Kopelman, TD Cowen

A: "in many cases, that means that a listing can sometimes be more expensive on Airbnb when it's cross-listed somewhere else… in making the migration, what we found is that many of the hosts did not take up their rate. Instead, the effective ADR to guests came down modestly… We are currently piloting in certain countries a further migration for our individual hosts, again, from the dual fee structure to the single service fee."
— Ellie Mertz, CFO

Assessment: This substantially revises the November read on the fee change. We characterized it then as trading a smaller monetization wedge for a lower shelf price, a deliberate concession. The disclosed outcome is better than that: hosts largely did not raise rates to preserve net earnings, so the guest-facing price fell without a corresponding hit to Airbnb's take, and the price-competitiveness problem being solved was specifically about cross-listed inventory appearing more expensive on Airbnb than elsewhere. That is a defect being repaired rather than margin being surrendered, which is why take rate is guided up in Q1 despite the migration expanding. The individual-host pilot is the item to watch, since that base is far larger and far less sophisticated about repricing.

Why Not Pursue Branded Hotel Chains

A question challenged the boutique-and-independent focus, arguing that branded chains would open more supply and bring more new users. The answer held the line on positioning while explicitly declining to rule chains out, and disclosed the New York supply figure.

Q: "And I get how the company was built, unique supply. But can you just talk about like why not lean into more brand hotels just because it could give the user and open up more supply and potentially bring in more new users to the platform?"
— Jed Kelly, Oppenheimer

A: "we think that we want to like just start with a huge number of boutiques and independents that are typically paying a higher commission than the chain and have been really aggressive with us reaching out saying that they would love to have another channel… We have more than 100 hotels in New York with more than 20,000 rooms available on the site just in New York City alone."
— Brian Chesky, Co-Founder & CEO

Assessment: The commission point is the analytically important one and it is easy to miss. Independent and boutique properties pay higher distribution commissions than chains, which have the scale to negotiate them down and the direct-booking programmes to bypass intermediaries entirely. By starting with independents, Airbnb is entering hotel distribution at the high-take-rate end rather than competing for the commoditized chain inventory where the incumbents' scale advantages are largest. The 20,000 New York rooms figure also shows the pilot is not a token: that is real inventory in the single most valuable urban market on the platform. "We're not saying where we're eventually going" is management leaving the chain door open, which is the correct optionality to retain.

What They're NOT Saying

1. What replaces the lapping product lift in H2 2026: management confirmed it will be comping the Q4 launches in the back half and said it intends to have incremental growth levers, without naming or sizing a single one. Given that the three disclosed features supplied roughly 300 basis points of Q4 GBV growth, this is the most consequential unquantified item in the FY26 guide.
2. Any revenue figure for experiences, services, or hotels: nine months after launch, the disclosure remains ratings, host counts, funnel attachment and now a nights-share band. Hotels were sized as a single-digit percentage of nights, which is progress; no revenue, commission rate, or take rate was given for any of the three.
3. The FY26 revenue growth number: the letter says at least low double digits and the CFO added on the call that this comes "with an ambition to grow even faster than that." That is a floor with an aspiration attached, not a range. With FX explicitly fading through the year, the difference between 11% and 13% is roughly $250M of revenue and is not resolvable from what was given.
4. Whether operations and support leverage resumes: the line grew 12.8% in Q4, faster than revenue, after declining in absolute dollars in Q3. Management described the AI deployment progress but did not address why the consolidated line lost leverage in the quarter, and did not commit to a level.
5. The conversion claim's denominator: chatbot-referred traffic converting better than Google-referred traffic is a striking assertion with no volume, no share of total traffic, and no trend attached. Without a denominator it is directionally interesting and financially unusable.
6. The hotel connectivity roadmap: asked directly about property-management-system integration friction and multi-rate support, management answered with the strategic case for hotels rather than an engineering commitment. That plumbing is the practical constraint on the pace of supply onboarding.
7. Loyalty, still: confirmed to be in active testing across different benefit tracks, with no launch window, for at least the third consecutive call. Sponsored listings remain explicitly gated on AI search, which is live to what management called a very small percentage of traffic.
8. The $2.0B current debt: reclassified to current liabilities on the December balance sheet and not addressed on the call or in the letter. Whether it is refinanced or repaid is a call on roughly a fifth of the corporate cash balance that competes directly with the buyback.
9. What "stable" margin means numerically: FY25 adjusted EBITDA margin was 35.1%. Stable could mean 34.6% or 35.6%, a range worth roughly $140M of EBITDA on the guided revenue base. Management has now committed to a direction without committing to a number.

Market Reaction

• Pre-print setup: ABNB closed at $115.96 on February 12, down 14.6% year-to-date against an S&P 500 that was itself down 0.2%, down 17.5% over the trailing twelve months, and down 17.2% over the trailing thirty days. The stock entered the print near the floor of its $105.69 to $161.42 52-week closing range, with sentiment as poor as it has been in our coverage.
• After-hours move: shares rose about 2% in extended trading on the initial release, which beat on revenue and adjusted EBITDA and missed on GAAP EPS.
• Next-day session (February 13): opened at $126.47, a 9.1% gap above the prior close, traded a $119.66 to $127.01 range, and closed at $121.35, up 4.6% or $5.39.
• Volume: 11.0M shares against a 4.2M thirty-day average, or 2.6x normal.
• Benchmark: the S&P 500 closed the session flat, so the entire move was idiosyncratic.

The gap between the roughly 2% immediate after-hours reaction and the 9.1% opening gap locates the repricing squarely in the earnings call rather than the press release. Three of the four things that moved this stock were call disclosures, not release disclosures.

The FY26 guide was the release's headline but the call's substance. The letter's outlook section carries the "at least low double digits" and "stable year-over-year" language, but the CFO's line-by-line explanation of where the incremental spend goes, and the CEO's framing of accelerating without needing billions of capital expenditure, are what made the guide credible rather than aspirational. A market that had marked this stock down 17% in thirty days was positioned for a margin-sacrifice guide and got the opposite.

The product-attribution disclosure cut both ways and net helped. Quantifying that three features supplied more than 200 basis points of nights and roughly 300 basis points of GBV growth tells a sceptic the acceleration is engineered and laps. It tells a bull that the growth algorithm is elastic to cheap product changes and that management can now measure and repeat the mechanism. On a stock trading near its 52-week floor, the second reading dominated.

The fade from $126.47 to $121.35 is the lapping problem being priced. Giving back roughly 40% of the opening gap on 2.6x volume is not a rejection of the quarter, which closed up 4.6%. It is the market discounting a guide whose first half is well-supported and whose second half depends on levers that have not been named. That is a reasonable discount and it is where the remaining upside sits.

Street Perspective

Debate: Does the FY26 Guide Survive the H2 Lapping?

Bull view: Management guided FY26 knowing exactly what laps and when, and disclosed the size of the lapping item voluntarily. The replacement levers are real and already in motion: Reserve Now, Pay Later going global and to US cross-border stays, the fee migration extending from property managers to individual hosts, incremental supply acquisition funded inside the flat-margin envelope, and expansion markets where India is compounding at 50% origin nights growth. AI search is in pilot with nothing in the outlook, so it is pure upside.

Bear view: Roughly 300 basis points of Q4 GBV growth came from three one-time product changes. Underlying nights growth was near 8%, essentially flat with Q3. The FY26 guide requires the underlying rate to improve while FX fades, and every named replacement lever is either a repeat of the same trick in a new geography or an unquantified investment. Q1's guided acceleration is itself substantially currency, Easter timing, and a booking-to-stay shift, none of which is demand.

Our take: The bear has the better description of Q1 and the bull has the better description of the year. Q1's 14–16% is genuinely flattered by roughly three points of currency plus about 50 basis points of Easter, so the underlying Q1 rate is closer to 11%. But 11% is still above FY25's 10.3%, and the replacement levers are the same mechanism applied to larger addressable bases: the fee migration's next tranche is the entire individual-host population against the property-manager subset, and Reserve Now, Pay Later's next tranche is every market outside the US against US domestic only. We model FY26 revenue at $13.6–13.8B and treat Q3 2026 nights growth as the decisive test.

Debate: Is Airbnb an AI Winner or an AI Casualty?

Bull view: Airbnb captures the entire cost benefit of AI with none of the capital cost. Roughly a third of North American support issues resolve without an agent today, heading to voice and to every language where the company has live agents, with more than 80% of engineers already using AI tools. There are no data centers and no models to train. Meanwhile assistant-referred traffic converts better than Google-referred traffic, so the new discovery layer is a channel improvement. The verified-identity, review, messaging, payments and insurance stack is not reproducible by a chat interface.

Bear view: Every travel supply set is ultimately addressable by an assistant that can transact, and the moat items named are integration work rather than genuine barriers. Airbnb declined the largest assistant integration when it was offered and is now arguing the channel is helpful, which is a change of position rather than a strategy. AI search is still in pilot with nothing in the outlook, and the support-cost leverage did not appear in Q4's consolidated operations and support line, which grew faster than revenue.

Our take: The cost-side case is proven enough to underwrite and the demand-side case is not yet proven either way. What tips it constructive is the capital asymmetry: in a market re-rating consumer platforms on AI capital intensity, a company that can improve unit costs without a capex cycle is structurally advantaged, and the CTO hire says management intends to press that advantage. The conversion claim needs a denominator before it belongs in a model. We credit the cost line, credit nothing to AI search, and treat the disintermediation risk as reduced from November but live.

Debate: Is 14x Trailing Free Cash Flow the Right Multiple for This Business?

Bull view: At $121.35 the shares carry roughly $65.5B of enterprise value against $4,613M of trailing free cash flow, a 14.2x multiple, and a 6.2% free cash flow yield on the equity. The company is guiding revenue to accelerate on flat margins, the tax rate is falling from 20% to the mid-to-high teens for structural reasons worth roughly $0.15 of EPS, the buyback consumed more than 80% of 2025 free cash flow, and there is $9.1B of net cash. The multiple sits a shade below the 15.0x at Q3, most of that difference coming from the lower share count rather than the price, while the outlook improved.

Bear view: GAAP earnings are $4.03 per share against $4.11 a year ago, so on the accounting the business went backwards in 2025. Free cash flow is flattered by float that grows with GBV, and full-year free cash flow margin already fell 271bp to 37.7%. Stock-based compensation of $1,592M grew 13.1%, faster than revenue, so a large share of the buyback is treadmill. Fourteen times free cash flow is not obviously cheap for a business that has grown revenue 10% for two consecutive years.

Our take: The GAAP point is the weakest of the bear arguments this quarter, because 2025's earnings decline is explained by the $213M CAMT charge in Q3 and roughly $90M of non-income tax matters in Q4, neither of which recurs, plus a tax rate that is about to fall 250 basis points or more. The stock-based compensation point stands and management has now committed to slower growth in it. The float point stands and is a genuine reason not to pay a premium multiple. But the setup is straightforwardly better than in November: same business, better guide, lower multiple, and a stock that spent the preceding month falling 17% into a print that beat. Fourteen times is the wrong multiple for accelerating revenue on stable margins.

Model Update & Valuation Framework

Valuation framework: At the $121.35 close and the Q4 diluted weighted-average share count of 614M, the equity is worth approximately $74.5B. Corporate cash and investments of $11,049M less $1,999M of current debt gives $9,050M of net cash, for an enterprise value of roughly $65.5B. Against FY25 revenue of $12,241M that is 5.3x sales; against FY25 free cash flow of $4,613M it is 14.2x; against FY25 adjusted EBITDA of $4,297M it is 15.2x. The free cash flow yield on the equity is 6.2%. Every one of those multiples is lower than at the November print despite a better forward outlook.

Why the base case is 16–17x: we held 16x at Q3 for a high-single-digit compounder with an unresolved margin question. The margin question is resolved and the growth rate is guided higher, which justifies the top of that band rather than a re-rating beyond it, because the H2 lapping risk is genuine and unaddressed. On $5.05B of FY26E free cash flow, 16x gives $80.8B of enterprise value and 17x gives $85.9B; adding $9.05B of net cash and dividing by 614M shares gives $146 and $155 respectively. The midpoint of $150 implies roughly 24% upside from $121.35.

Risk-reward: base $150 (+23.6%), bull $189 (+55.7%), bear $108 (−11.0%). The base-to-bear ratio is roughly 2.1 to 1 and the bull-to-bear span is roughly 4.4 to 1, both materially better than the 0.7 to 1 and 2.2 to 1 we computed in November at a marginally lower share price. The bear case now requires the H2 lapping to bite hard enough to take revenue growth back to 8% and the margin guide to break, which is two failures rather than one. That asymmetry supports Outperform.

Thesis Scorecard Post-Earnings

Overall: Thesis strengthened. Two of the three bear points from the initiation moved from Emerging to Contained on this print, and both moved for the right reason: management made a forward commitment rather than delivering a favourable one-quarter outcome. The bear case has not disappeared, it has changed shape. It was margins compressing and monetization eroding; it is now the narrower and more testable proposition that the engineered acceleration laps in H2 2026 with nothing named to replace it. That is a better bear case to underwrite against because it has a date attached.

Action: Upgrading to Outperform from Hold. The specific upgrade trigger set in the November note, a 2026 adjusted EBITDA margin guide at or above 35% with nights holding mid-to-high single digits, was met in full and accompanied by a revenue acceleration guide and a Q1 take-rate expansion we had not assumed. The multiple eased to 14.2x trailing free cash flow, mostly on the lower share count rather than the price, while the outlook improved. We would downgrade back to Hold on any of: Q2 or Q3 2026 nights growth falling below 7%, a mid-year cut to the stable-margin guide, or evidence that the individual-host fee migration behaves differently from the property-manager cohort. We would consider Hold on valuation alone above roughly $175 absent an estimate revision.