Core Markets Accelerate and the Guide Goes Up a Third Time, But a 17% Session Took the Upside With It

Results vs. Consensus

Q2 2026 Scorecard

Year-Over-Year Comparisons

First-Half 2026 vs. 2025

Cross-check: the two reported 2026 quarters sum to the six-month figures. Revenue $2,678M + $3,608M = $6,286M. Net income $160M + $816M = $976M. Adjusted EBITDA $519M + $1,261M = $1,780M. Free cash flow $1,704M + $1,253M = $2,957M. Nights 156.2M + 148.3M = 304.5M. Stock-based compensation $410M + $487M = $897M. No reconciling items.

Quality of the Print

Revenue: The 16.5% headline is a point slower than Q1's 17.9%, and every layer beneath it is better. Ex-FX revenue growth was 13% against 15% in Q1, but ex-FX GBV growth was 15% against 13%, a two-point acceleration that management called out explicitly. That is the metric to watch, because GBV leads revenue by the booking-to-stay interval and Reserve Now, Pay Later has been lengthening that interval. For the first time in our coverage the underlying booking line accelerated on a currency-neutral basis rather than merely appearing to on a reported one.

Margins: The 124 basis points of operating margin expansion decompose the same way they did in Q1, and the components sum. Sales and marketing added 193 basis points of cost-ratio pressure while growing 26.6% against 16.5% revenue growth. Against that, general and administrative gave back 135 basis points on essentially flat dollars, product development 108, operations and support 72, and cost of revenue 3. The CFO's characterization matches the arithmetic exactly: expansion driven by revenue growth and efficiency in operations and support and product development, partly offset by continued sales and marketing investment. Adjusted EBITDA margin of 35.0% against 33.7% is the second consecutive year-over-year expansion after two quarters without one.

Free cash flow: The line that broke in Q1 has partly healed. Free cash flow of $1,253M grew 30.2% against 16.5% revenue growth, and the margin rose to 34.7% from 31.1%. Trailing-twelve-month free cash flow of $4,827M grew 12.9% and the trailing margin ticked back up to 37% from 36% at Q1. But the underlying float dynamic is unchanged: unearned fees of $2,831M were slightly below the $2,857M of a year earlier despite GBV growing 15.7%, and management repeated that absent Reserve Now, Pay Later, unearned fees would have grown. The Q2 recovery is operating-income-driven, not float-driven, which is the right way for it to happen and confirms the valuation-anchor change we made in May.

Key Business Metrics

Regional Performance

North America — the best growth in almost three years

North America held high-single-digit nights growth for a second consecutive quarter and management characterized it as the strongest in nearly three years, attributing it to strong underlying travel demand rather than to a single feature. ADR rose 7% on price appreciation and mix, with four-bedroom-plus entire homes continuing to outgrow long-term stays and private rooms, a mix shift now more than a year old.

Assessment: This is the region that decides whether the mid-teens guide holds, and it is the one we flagged in May as most exposed to the second-half lapping of Reserve Now, Pay Later. A second consecutive high-single-digit quarter, with the CEO naming the US among four core markets that accelerated, is the strongest evidence yet that the improvement outlives the feature that started it. The ADR mix point matters more than it reads: bedroom nights growing faster than nights means the ADR gain reflects guests buying more house rather than paying more for the same house, which is a durable form of price growth.

EMEA — the conflict drag proved smaller than feared

EMEA reaccelerated from mid-single to high-single-digit nights growth as the Middle East conflict headwind faded, and the CFO said the impact on the business was less than anticipated. The Q3 guide assumes no significant conflict impact at all, against the roughly 100 basis point allowance carried in the Q2 guide.

Assessment: The substitution argument we credited in May held up in practice. A geographically diversified marketplace reroutes disrupted demand rather than losing it, and the recovery came faster than the company's own planning assumption. The forward risk is now asymmetric in the other direction: with zero conflict drag assumed in Q3, any renewed disruption is an unguided negative rather than a cushioned one.

Latin America and Asia Pacific — the expansion cohorts keep compounding

Latin America accelerated to roughly 20% nights growth, with Brazil origin net nights accelerating past 30% and first-time bookers up 40%, and Mexico supported by installment payments introduced in June. Asia Pacific held high-teens growth, with India origin net nights accelerating to 60% and first-time bookers there more than doubling. Management said it continues to see broad market-share gains across Latin America.

Assessment: Brazil accelerating to over 30% origin nights growth, three years into being a priority market and now a top-five contributor, is the strongest single vindication of the country-by-country playbook in the whole coverage period. It answers the obvious objection that expansion markets decelerate as they mature. India at 60% with first-time bookers doubling is the same story two years earlier in the cycle. The one caveat is ADR: Latin America's 9% ADR gain was almost entirely currency (only 2% ex-FX) and Asia Pacific's was 1%, so these regions add volume and TAM rather than revenue mix.

Supply

Supply strategy remains explicitly selective rather than maximal, targeted at markets where demand already exists and supplemented by pushing existing hosts to open more calendar availability. The World Cup channel delivered: more than 150,000 listings from first-time hosts across the 16 host cities since outreach began in October 2025, against the 100,000 disclosed at the Q1 print. Airbnb now reports more than 5.5 million hosts, up from more than 5 million a year ago.

Assessment: The World Cup cohort came in 50% above the figure management gave three months ago, which is the sort of beat that gets no coverage and compounds for years. On the Paris retention rate management disclosed in May, roughly half of those 150,000 listings should persist, which is 75,000 durable listings in exactly the North American markets where supply density most constrains conversion. That is the supply-side equivalent of the core-market demand acceleration, and the two together are why the second-half guide is more defensible than we assumed.

Key Topics & Management Commentary

Overall Management Tone: The most expansive call of the four we have covered, and the first in which management actively resisted attributing results to any single driver. The CEO spent the prepared remarks arguing that the quarter reflects an operating-model change rather than a product win, and returned to that framing three times in Q&A. Confidence extended well beyond the guide, into a three-phase expansion vision reaching past travel and an unprompted acknowledgment that the company is sitting on cash and sees M&A opportunities. Where the tone runs ahead of disclosure is on the newer businesses, where superlatives (best hotel booking product online, extraordinary AI search tests) arrive without a single revenue figure attached.

1. Core Markets Accelerated, Which Changes the Shape of the Story

The structural news of the quarter is not the aggregate growth rate but where it came from. For four quarters the pattern has been expansion markets growing at roughly twice the core rate while core markets sat flat. That broke.

"What's especially encouraging is that we're not only seeing growth accelerate in our emerging markets, we are seeing growth accelerate in many of our core markets. In fact, the U.S., France, U.K., and Australia all accelerated in growth in Q2." — Brian Chesky, Co-Founder & CEO

Nights growth was high-single-digit in both North America and EMEA, roughly 20% in Latin America, and high teens in Asia Pacific. North America's rate was the best in almost three years.

Assessment: The single most consequential disclosure in the release and the one the market repriced on. Core-market acceleration is worth more than an equivalent amount of expansion-market growth for three reasons: the ADR is materially higher, the take rate is established rather than developing, and it removes the objection that Airbnb's headline growth depends on low-monetization geographies. It also makes the second-half guide credible without leaning on currency, which is the concern we raised in May. This is the datapoint that would have made us more constructive at a lower price.

2. The Third Consecutive Guide Raise

FY26 revenue growth was raised from low-to-mid teens to at least mid teens, and the adjusted EBITDA margin floor from at least 35% to at least 35.5%. That is the third raise in three quarters, against an original February guide of at least low double digits with a stable margin.

"We now expect year-over-year revenue growth to improve to at least mid-teens, up from the low-to-mid-teens guidance we provided last quarter, supported by the accelerated pace of nights and seats booked we've observed across our business." — Ellie Mertz, CFO

Q3 revenue is guided to $4.69–4.77B with roughly three points of currency tailwind, GBV growth in the mid teens, and nights growth in the low double digits, which would be the first double-digit nights quarter of our coverage. Q3 adjusted EBITDA is guided up year-over-year with margin down slightly on investment timing.

Assessment: Three raises in three quarters on a guide originally set in February is an unusually strong execution record and it is the reason this downgrade is a valuation call rather than a business call. Two details temper it. The Q3 revenue guide's low end computes to +14.5% against the reported Q3 2025 revenue of $4,095M, so the stated 15 to 17% range rounds up at the bottom. And roughly three points of the Q3 guide is currency again, the third consecutive quarter of a similar tailwind, which will eventually lap.

3. The Take-Rate Walk-Back Hidden Inside the Raise

In May management said the migration to a single fee structure and the insurance programme were expected to lift the full-year take rate. This quarter the full-year take rate is guided relatively flat, and the reason is new.

"For the full year, we expect our implied take rate to be relatively flat compared to 2025, accounting for the timing of bookings versus check-in with Reserve Now, Pay Later, as well as higher customer incentives related to new businesses during 2026. Absent these incentives, we would have anticipated our implied take rate to be slightly higher during the year, driven by our monetization initiatives and execution across our product roadmap." — Ellie Mertz, CFO

The incentives are visible in the release. Featured hotels carry a price match guarantee and up to 15% Airbnb credit toward a guest's next booking, with the credit offer running to the end of 2026. Separately, the CEO described Airbnb as offering hotels an extremely favourable take rate as part of the supply pitch.

Assessment: The most important thing in this release that nobody will write about. Three quarters of adjacency enthusiasm have been costless in the reported numbers because the adjacencies were too small to matter; this is the first quarter in which their cost lands in a guided metric. Two separate levers are at work: guest-side incentives (the credit and the price match) and supply-side pricing (a deliberately low commission to win hotel inventory). Both are correct customer-acquisition spending, both are being funded out of the take rate rather than out of sales and marketing, and neither is quantified. Investors modelling the adjacencies as pure upside should note that the company has just told them the near-term cost shows up in monetization.

4. Hotels Inflected, and Management Is Stepping on the Gas

Hotel nights are growing roughly three times as fast as the homes business, supply spans more than 20 top destinations including New York, Paris, London, Madrid, Rome and Singapore, and the strategic constraint management named in May (execution speed) appears to have been resolved. The CEO's framing was unusually personal.

"The hotel initiative is going significantly better than I expected. I had high expectations… That's where my expectations were not high enough, because after we began outreach for hotels, things flipped. What we are now seeing is a huge amount of influx of interest from hotels wanting to list on Airbnb." — Brian Chesky, Co-Founder & CEO

The critical operational update is that hotels are working in unconstrained markets, not just in the regulatory-constrained cities where the pilot began. Roughly 35% of first-time hotel guests return to book a home.

Assessment: Working in unconstrained markets is the finding that converts hotels from a supply-gap product into a genuine second business, because it means the demand is being created rather than merely recaptured. Note the disclosure change on the conversion metric though: May's figure was roughly 55% of guests who book a hotel returning to book a home; this quarter's is roughly 35% of first-time hotel guests. Those are different populations, and the newer, narrower cut is the more useful one for underwriting incremental acquisition. It is also the lower one, and a reader comparing the two headline numbers without reading the definitions would draw the wrong conclusion about the trend.

5. The Single Service Fee Goes Universal by Year-End

The fee migration that began with property-management-software hosts in October 2025 is being extended to essentially the whole host base. Approximately half of active listings are already on the single 15.5% service fee, against more than a quarter at the Q1 print, and management expects the remainder to complete this year.

"we did extensive testing on a broader swath of hosts this winter and spring and have begun migrating the remainder of our hosts to the single service fee. We anticipate by year-end, our entire supply base will be on that single service fee… that single service fee allows us to provide more simplified pricing recommendations and in aggregate, has a kind of downward pressure on pricing to allow us, one, to offer more value to guests, but also ensure that we are priced competitively relative to other platforms." — Ellie Mertz, CFO

Assessment: The scale of this is easy to underestimate. Every listing on the platform will have changed its fee architecture within roughly fifteen months, and management is explicit that the aggregate effect is downward pressure on displayed prices. That is a deliberate competitive repricing of the entire marketplace against hotels and the incumbent agencies, executed without a headline. It is also the largest untested variable in the model: the property-manager cohort repriced rationally because pricing is their job, and individual hosts are less likely to. The Q3 and Q4 take-rate and ADR lines are where any friction will show.

6. AI as the Operating Model, Not a Product

The CEO's central argument was that the quarter reflects a change in how the company builds rather than what it built. Two metrics support it: concept-to-launch time reduced by as much as 60% on key initiatives, and features and improvements shipped up nearly 80% against the same six months a year ago.

"We've rebuilt Airbnb from the ground up to be an AI-native company. We have more world-class talent than ever before. AI is transforming how we execute and build products. Said simply, AI is the best thing to ever happen to Airbnb." — Brian Chesky, Co-Founder & CEO

On customer support, the AI assistant now runs in more than 50 languages, resolves nearly 45% of the issues it opens without a human agent (against more than 40% in Q1 and roughly a third in Q4), and was benchmarked by a third-party consulting firm as the best AI support assistant among six major US travel platforms. Voice support begins rolling out later this year. Customer-support cost per booking fell about 16% year-over-year, against 10% in Q1.

Assessment: The velocity metrics are the ones we would put weight on, because they are the mechanism behind the compounding-improvements thesis that has driven three guide raises. Eighty percent more shipped features at 60% faster cycle time is a genuine change in the rate at which a product organization can test hypotheses, and it explains why the company keeps finding new growth levers rather than exhausting them. The support metric is now a proven, accelerating cost deflator: from 10% to 16% cost-per-booking reduction in one quarter, with voice still ahead.

7. The AI Cost Question, Answered With Unusual Specificity

Asked how the AI-native transition affects product costs, the CEO made an argument about consumer AI economics rather than about Airbnb's budget: nearly four years after ChatGPT launched, most AI businesses are enterprise because consumer inference costs are hard to cover, and Airbnb is an exception because its transaction values are large.

"the inference costs of Airbnb are de minimis relative to the ROI of our business model. Right? We're not in the business of information where we're trying to monetize. Our transactions are very high-dollar transactions, if AI can just increase our conversion rate just a little bit, the inference cost is so outweighed by the amount of money we make on that increased ROI." — Brian Chesky, Co-Founder & CEO

The CFO added the number that matters for the model, and the two offsets.

"in the updated guidance that we provided, it obviously does assume a material increase in terms of the AI spend over the course of the year. I would note that, yes, we are expanding margins while absorbing that increased cost… what we're seeing is that we don't need to grow our head count at levels that we did in the past because we're getting so much more output and speed from our existing workforce." — Ellie Mertz, CFO

Assessment: This resolves the tension we flagged in May between the CEO saying AI would not be visible in the P&L and the CFO naming AI as a growing reinvestment bucket. The reconciliation is that AI spend is rising materially and is being more than offset by support-cost deflation and headcount restraint, which is why the margin expanded anyway. The high-ticket argument is also the correct structural point: a company earning roughly $24 of revenue per night booked can afford inference costs that a company monetizing attention cannot. This remains the cleanest AI setup in consumer internet and it is now demonstrated in the margin rather than asserted.

8. Host Pricing Tools Named as a Bigger Lever Than Reserve Now, Pay Later

Asked about the new personalized host recommendations, the CEO made the most consequential forward statement on the call, sizing the pricing workstream against the feature that has driven the last four quarters of acceleration.

"I think this is one of the biggest single levers for growth that we have. I think it's significantly greater than Reserve Now, Pay Later. If you want to just put it in perspective, it is, I don't know, I don't want to say a multiple, but many multiples bigger than RNPL." — Brian Chesky, Co-Founder & CEO

The mechanism is an AI pricing model ingesting hotel prices, Airbnb prices, local events and lead-time behaviour, surfaced to hosts as one-tap recommendations with day-level dynamic pricing, alongside the calendar.

Assessment: If this is directionally right it is the answer to the lapping question we have carried since February. Reserve Now, Pay Later plus the cancellation-policy and fee changes contributed roughly 3 points of nights growth and 4 points of GBV growth in Q1; a lever many multiples larger would more than replace them. The claim is also unfalsifiable at this stage and management has an obvious incentive to point at the next thing when the last thing is about to lap. We would treat it as a genuine option with no value in the model until the Q4 or Q1 print shows a quantified contribution, exactly as management quantified the previous three features.

9. AI Search Finally Enters Test, Behind a Toggle

After three quarters of being described as in pilot with nothing in the outlook, AI search is entering live test this month at a small share of traffic, expanding through the year. The rollout design is deliberately conservative: existing search remains the default with a toggle above it, mirroring the approach used for the total-price-display change.

"people come to Airbnb, often, most people, a handful of times a year, and they have an expectation that they see a search box with a location. It's going to take some time, months and months, to retrain the customer. The way we're initially going to roll it out is the default is going to still be the core search. Above, you'll see a toggle." — Brian Chesky, Co-Founder & CEO

The described feature set spans natural-language input, natural-language response, AI-generated conversational listing titles, and real-time personalized listing-page highlights, with the CEO expecting materially higher conversion for users who opt in.

Assessment: The toggle is the right decision and it is also the reason this should stay at zero in any 2026 model. A default-off feature at a small share of traffic on a platform where users transact a handful of times a year cannot move a full-year number, and management has consistently said nothing is baked in. The strategic significance is different from the financial one: shipping a working conversational search in a category the CEO argued in May that nobody had solved would be the strongest possible answer to the disintermediation question. That is a 2027 story.

10. Bedroom Nights and the Durable-ADR Argument

A new metric was introduced: Bedroom Nights Booked, defined as nights multiplied by bedroom count, which grew more than 12% against nights growth of 10%, with more than a billion bedroom nights booked over the trailing twelve months, a record.

"we have been aggressively pushing pricing tips and affordability across the platform, and you've seen yet the nominal ADRs rise over that period… What we've seen is that the growth in what we call bedroom nights has, one, been at a higher level, and also accelerated more on a year-over-year basis than our nights booked… that component of ADR appreciation that is durable and really a reflection of incremental value delivered, not just rising prices" — Ellie Mertz, CFO

Assessment: A genuinely useful disclosure and a fair rebuttal to the objection that Airbnb's ADR growth contradicts its affordability push. If bedroom nights outgrow nights, guests are consuming more accommodation rather than paying more per unit of it, which makes the ADR line a mix effect rather than an inflation effect. Two caveats: the metric is new and therefore unaudited by any history, and a company introducing a favourable new metric in the same quarter its stock re-rates 17% deserves the usual scepticism. We will track it, and we would want to see it disclosed in a down quarter before treating it as a permanent part of the framework.

11. Services, and the Three-Phase Expansion Vision

The Summer Release expanded services to grocery delivery, car rentals, airport pickups and luggage storage, with resort passes added recently. Car rentals is expected to be the largest by revenue, and management was surprised that rental durations exceed the average length of an Airbnb stay. Development cycles are compressing sharply: groceries took eight to nine months, airport pickups about six weeks. Most are partnerships, so fulfilment costs sit with the partner.

The CEO laid out a three-phase framing: attaching services to home-intent traffic today, becoming a one-stop shop for travel next, then extending from travelling to living and eventually to other ways for people to connect.

"I do imagine that in the near term, being over the next year and a half, you'll absolutely be able to get onto Airbnb with a broad travel intent, like, 'I'd like to travel with my family to somewhere in Europe. Can you recommend where I should travel? Can you put my whole trip together?'" — Brian Chesky, Co-Founder & CEO

Assessment: The partnership structure is the important economic detail: lead generation with the fulfilment cost borne by the partner is a high-margin, low-capital way to widen the offer, and the compressing development cycle (nine months to six weeks) is the same velocity story showing up in a different place. The three-phase vision is where we would apply the most discount. Phase one is happening and measurable; phase two is a real 2027 product question; phase three is a founder describing ambition, not a plan. Note also the M&A signal, unprompted: management described its strategy as opportunistic while pointing out that it holds substantial cash and that entrepreneurs would want Airbnb stock. That is a company preparing its shareholders for something.

Guidance & Outlook

Testing the FY26 raise: H1 2026 revenue of $6,286M grew 17.1%. Adding the Q3 guide midpoint of $4,730M gives nine-month revenue of $11,016M against $9,463M, or +16.4%. At least mid-teens for the full year implies roughly $14.1–14.2B, which requires Q4 2026 revenue of approximately $3.12B against $2,778M, or about +12%. The guide therefore still embeds a decelerating fourth quarter against the toughest comparison of the year, which is the same disciplined construction we credited in May. The raise is not an extrapolation.

Testing the margin floor: H1 adjusted EBITDA of $1,780M is a 28.3% margin. A full-year floor of 35.5% on roughly $14.15B implies about $5.02B, so the second half needs approximately $3.24B on roughly $7.86B of revenue, or a 41.2% margin. The second half of 2025 delivered 41.3% on the same seasonal shape. The floor therefore requires second-half margin to be flat year-over-year, which is consistent with the Q3 guide of margin down slightly and implies a stronger Q4. Achievable, and not conservative.

The Middle East cushion is gone. The Q2 guide carried roughly 100 basis points of assumed conflict drag on nights; the actual impact was smaller than anticipated. The Q3 guide assumes no significant impact at all. That is a reasonable call given the observed recovery in EMEA, and it removes the buffer that made the Q2 nights guide easy to clear. Any renewed disruption in Q3 lands directly on the guided number.

What the take-rate change costs: holding FY26 implied take rate flat at FY25's 13.4% on roughly $14.15B of revenue implies GBV of approximately $105.6B, or +15.7%. Had the take rate landed slightly higher as guided in May, say at 13.6%, the same GBV would have produced roughly $200M more revenue. Investors should read the new-business incentive spend as roughly that order of magnitude, funded through monetization rather than through the sales and marketing line where it would be visible.

Analyst Q&A Highlights

Whether Hotels Improve Conversion in Markets That Are Not Supply-Constrained

The opening question went to the sharpest test of the hotel strategy: the supply-gap logic obviously works where regulation caps home supply, but does adding hotels lift conversion in cities where Airbnb already has plenty of inventory? The answer was unequivocal and included the most candid line of the call.

Q: "Can you tell us how the initiative's going and what you're seeing in maybe the cities where you have a lot of inventory, not just the cities where you might have had some regulatory restrictions, but is it improving conversion rates in cities where you already have a lot of inventory?"
— Justin Post, Bank of America

A: "The hotel initiative is going significantly better than I expected… we are seeing strength in both supply-constrained markets and in non-supply-constrained markets… The basic theory of our marketplace is as we add more supply of homes or hotels, conversion rate of our traffic goes up and we see more bookings… about 35% of people who come to Airbnb and book a hotel for the first time, come back and book a home."
— Brian Chesky, Co-Founder & CEO

Assessment: The question isolated exactly the right variable and the answer changes the size of the opportunity. If hotels lift conversion only where home supply is capped, the ceiling is a handful of regulated cities. If they lift it everywhere, the ceiling is the entire traffic base, and the CEO's separate observation that Airbnb's conversion rate is significantly below the incumbents' becomes the sizing framework. The offsetting detail is buried in the same answer: winning that supply involves an extremely favourable take rate for hotels, which is where part of the guided take-rate flattening comes from. Higher conversion at a lower rate is still accretive, but it is not the pure upside the enthusiasm implies.

Whether the Ancillary Products Are Attach Rates or a Full-Trip Ambition

A question pressed on the end state for services: are these attachments to an existing home booking, or is the ambition a full-service travel platform that fulfils an entire trip from a single natural-language request? The answer laid out an explicit three-phase roadmap and volunteered an unusually direct statement on M&A capacity.

Q: "Are these simply sort of attach rates, someone books a home, and then we'll add on car hire and other stuff. Or could you imagine a world where you go to Airbnb and say, 'Please book me a two-week trip to Scotland,' and it will be able to fulfill the whole trip, and be a sort of full service travel OTA in that regard?"
— Richard Clarke, Bernstein

A: "The next stage has become a one-stop shop for travel. You'll eventually be able to get all of your travel needs through Airbnb, and that's what you're starting to see. Phase II would be we will go from traveling to living. Phase III, we go from traveling to living to other ways for people to connect on our platform… Our M&A strategy remains opportunistic. As we go into a lot of new verticals, there's a lot of M&A opportunities. We are sitting on quite a lot of cash."
— Brian Chesky, Co-Founder & CEO

Assessment: Two things to take from this. The near-term one is that phase one, attaching services to existing home intent, is where essentially all of the current contribution sits, which management stated plainly. The forward one is the M&A language, which was volunteered rather than extracted and is a change in posture from a company that has bought almost nothing since HotelTonight. An investment-grade balance sheet with roughly $9.7B of net cash, a stated interest in new verticals, and a founder observing that entrepreneurs would want the stock is a set of facts investors should hold together.

Hotel Scaling Detail and When AI Search Becomes a Real Part of the Product

A two-part question asked for concrete evidence on hotel conversion improvement and scale, and for the status of AI search including whether it gates an advertising product. The hotel answer was the more disciplined of the two; the AI search answer set out the rollout mechanics for the first time.

Q: "Maybe you can give us a sense of maybe anecdotes on markets where you're seeing particular conversion rate improvement or areas where you're comprising a meaningful portion of hotels bookings… The second one, just wondering how the AI search tests are going. When you expect that to be a bigger part of the product, and if that's a gating factor to an ad product, what you think about that."
— Lloyd Walmsley, Mizuho

A: "we've shared that today, hotels are only a single-digit % of nights booked on the platform, so a relatively small segment. As we disclosed in the letter, given our efforts over the last several months, hotels, that segment, is currently growing about three times as quickly as homes."
— Ellie Mertz, CFO

Assessment: Note what was not given. The question asked for market-level conversion evidence and got a platform-level growth ratio, which is a different and less falsifiable claim. Four quarters into the hotel programme the disclosure set is still a nights-share band, a relative growth rate, and a conversion ratio whose definition changed this quarter. The advertising half of the question went unanswered entirely, which is now the third consecutive call on which sponsored listings have been raised and deferred behind AI search.

Whether the Single Fee Structure Extends to Every Host

A short, precise question asked whether the fee simplification stays with professional property managers or reaches the entire host base. The answer confirmed a platform-wide migration completing this year and, notably, stated the pricing intent out loud.

Q: "Just following up on the changes in fees, are you planning to roll that out to every host, or is that mostly still geared towards professional property managers?"
— Jed Kelly, Oppenheimer

A: "We anticipate by year-end, our entire supply base will be on that single service fee. As a reminder, that single service fee allows us to provide more simplified pricing recommendations and in aggregate, has a kind of downward pressure on pricing to allow us, one, to offer more value to guests, but also ensure that we are priced competitively relative to other platforms."
— Ellie Mertz, CFO

Assessment: The most operationally significant answer of the call and the one carrying the most execution risk into the second half. Repricing the entire supply base within fifteen months, with the explicit intent of pushing displayed prices down, is a marketplace-wide intervention. The property-manager cohort absorbed it cleanly because repricing is their profession; individual hosts are a far larger, far less sophisticated population, and the company is relying on them to adjust nightly rates to preserve their own net earnings. If a material share simply does not, either host earnings fall (a supply-retention risk) or guest prices do not (removing the competitive benefit). This is the variable we would watch hardest in Q3 and Q4.

How Much Incremental Margin Gets Reinvested Versus Dropped Through

A question aimed past 2026 to the structural question of long-run incremental margins: after a multi-year investment cycle now producing demand, how much of the incremental margin has to be reinvested to sustain momentum? Management declined to guide and instead pointed at the track record as the answer.

Q: "Can you talk a little bit about the interplay between incremental margins in the business over the long term, and how much of that incremental margin you think needs to be reinvested back into the operating momentum, as opposed to how much of that incremental margin might drop to the bottom line?"
— Eric Sheridan, Goldman Sachs

A: "last year we did have a bit of an investment cycle. Given the growth that it helped spawn for the current year, we're able to both accelerate growth Invest substantially behind that growth and also deliver an expansion in the margin… given the track record and the somewhat steady EBITDA margins that we have delivered, I think you can see there's a relative floor in our ability to continue to invest against that."
— Ellie Mertz, CFO

Assessment: The honest translation is that incremental margin will keep being spent for as long as growth opportunities keep appearing, and the 35%-plus adjusted EBITDA margin is the constraint the spending is fitted to rather than a floor beneath which earnings compound. That is the right operating posture for a company that has just re-accelerated, and it is also why nobody should model margin expansion beyond the guided floor. For a stock now trading at roughly 20.6x our FY26 adjusted EBITDA estimate, the absence of a margin-expansion path beyond 35.5% is a meaningful part of why the upside has compressed.

How AI Pricing Tools Work, and Whether Experiences Move the Nights Line Yet

A two-part question asked how the new host recommendation tools help set optimal prices, and whether experiences have reached a scale that contributes to reported nights and seats. The pricing answer produced the call's biggest forward claim; the experiences answer was refreshingly deflationary.

Q: "I want to hear a little bit more about the pricing side and how Airbnb is helping hosts sort of decide what the optimal pricing is for each home that's available. The second question is on experiences… whether experiences are at a size and scale yet to actually contribute to overall nights and seats booked."
— Ron Josey, Citi

A: "AI is able to take in a lot of data sources. We're able to take in a lot of data sources of hotel prices, of Airbnb prices, events coming to town, the nature of lead time bookings, and based on that, we have a new pricing model that we're rolling out to hosts… With regards to experiences, they're growing really quickly, but they're growing on a small base."
— Brian Chesky, Co-Founder & CEO

Assessment: The pricing tool is the most credible candidate to replace the lapping product lift, and the mechanism described (hotel prices, competitor rates, local events, lead-time behaviour, one-tap adoption) is a genuine capability rather than a slogan. The value is that it attacks the affordability gap versus hotels, which the CEO has named as the core constraint on the business for four consecutive calls. On experiences, the answer was appropriately cold: small base, multi-year horizon, and an explicit ranking that puts homes first, hotels and international expansion second, and services and experiences last. Investors should adopt that ranking.

What the AI-Native Transition Does to Product Costs

A question asked directly what the shift to AI-native development does to the cost base and what operational adjustments are being made to contain it. The answer was a broader argument about why consumer AI has been hard to monetize and why Airbnb is structurally exempt.

Q: "With AI helping drive faster velocity of innovations, I was hoping to get your perspective on how this transition to AI native could impact your product costs and if there are any operational adjustments you're making to help minimize the impact."
— John Colantuoni, Jefferies

A: "we do not need to make any major capital investments. We are not buying up a whole bunch of GPUs. Second, the inference costs of Airbnb are de minimis relative to the ROI of our business model… We're not so-called token maxing, which I think is this thing where I think all these CEOs at the beginning of the year have this mandate. 'I want to see everyone use AI,' with a vanity of have people use as many tokens as possible."
— Brian Chesky, Co-Founder & CEO

Assessment: The most differentiated AI answer we have heard from a consumer platform, and the most checkable. The claim is that AI is a net cost reducer here rather than a net cost, and the evidence supports it: support cost per booking down 16%, headcount growth explicitly moderating, margin expanding while the CFO confirms a material increase in AI spend. The token-maxing aside is also a genuine governance signal, since the failure mode for AI-native reorganizations is measuring adoption rather than output. That said, this argument is now fully priced. It was an underappreciated asymmetry in November; it is a widely held view in August.

What They're NOT Saying

1. How large the new-business customer incentives are: disclosed as the reason the full-year take rate went from a guided lift to guided flat, with no dollar figure, no rate, and no duration beyond the December 31 2026 expiry on the hotel credit offer. On our arithmetic the swing is on the order of $200M of revenue.
2. Any revenue figure for hotels, experiences or services: five consecutive quarters of nights-share bands, relative growth ratios and conversion percentages with no dollars attached to any of the three.
3. Market-level hotel conversion evidence: asked for directly and answered with a platform-level growth ratio. Four quarters into the programme there is still no city, no room-night count, and no commission rate.
4. Why the hotel-to-home conversion metric changed definition: May's figure was roughly 55% of hotel bookers returning to book a home; August's is roughly 35% of first-time hotel guests, measured over a specific July 2024 to June 2025 cohort window. Both may be accurate; the change was not explained.
5. Any quantification of the host-pricing lever: described as many multiples bigger than Reserve Now, Pay Later, which management has quantified at roughly 3 points of nights growth. A lever of that stated size with no sizing, no timeline and no rollout share is the single largest unquantified claim in the release.
6. Sponsored listings, for a third consecutive call: raised again, deferred behind AI search again, with no framework and no timeline.
7. Loyalty, for a fifth consecutive call: not mentioned at all this quarter, having been described as on the table in each of the previous four.
8. What happens if individual hosts do not reprice: the entire supply base moves to the single service fee by year-end with an explicit intent to push displayed prices down. Management did not address the case where individual hosts fail to adjust and simply absorb lower net earnings.
9. 2027 in any form: asked directly about long-run incremental margins and declined, pointing at the track record instead. With the stock re-rated on a mid-teens growth guide, the absence of any 2027 framework is more material than it was a quarter ago.
10. What the M&A cash might be for: volunteered that the strategy is opportunistic, the balance sheet holds substantial cash, and new verticals present opportunities. No category, no size, no criteria.

Market Reaction

• Pre-print setup: ABNB closed at $151.64 on August 6, up 11.7% year-to-date against the S&P 500's +12.6%, up 16.2% over the trailing twelve months, and up just 1.9% over the trailing thirty days. The stock sat within 1% of the top of its $111.54 to $153.11 52-week closing range, having gone sideways for three months after the May print.
• Next-day session (August 7): opened at $164.70, an 8.6% gap above the prior close, traded a $163.45 to $178.48 range, and closed at $178.07, up 17.4% or $26.43. The close came within $0.41 of the intraday high.
• Volume: 15.9M shares against a 3.5M thirty-day average, or 4.5x normal, by a wide margin the heaviest session of our coverage.
• Benchmark: the S&P 500 rose 0.6%, so essentially the entire 17.4% move was idiosyncratic.
• Level: the close sits roughly 16% above the pre-print 52-week closing-range top.

The defining feature of this session is the absence of a fade. In February the stock gapped 9.1% and gave back roughly 40% of it by the close. In May it gapped down 1.7%, rallied to a new high intraday, then surrendered most of the gain. Here it opened up 8.6%, doubled the gap through the day, and closed within forty-one cents of the high. Sellers never showed up.

Core-market acceleration is what got repriced. A 17.4% move on a 0.8% revenue beat is not a reaction to the print. It is a reaction to a change in what the business is understood to be. The market had been paying for a high-single-digit core growing slowly while expansion markets compounded; it was shown four core markets accelerating simultaneously and a Q3 nights guide in the low double digits, and it re-rated the growth algorithm rather than the quarter.

The third consecutive raise did the rest. Guidance that has moved from at least low double digits with a stable margin in February to at least mid teens with at least a 35.5% margin in August, without a single walk-back on either line, establishes a credibility that a single quarter cannot. The market is now underwriting management's guide rather than discounting it, which is worth multiple points on its own.

And the positioning was clean. Up 1.9% over thirty days into the print, with the stock having done nothing since May, this was not a crowded long. That absence of froth is precisely why the move ran so far in one session, and it is also why we do not expect a give-back: this looks like new money establishing positions rather than existing holders adding.

Street Perspective

Debate: Is the Core-Market Acceleration Durable or a Comp and Currency Artifact?

Bull view: Four core markets accelerated simultaneously, North America posted its best nights growth in almost three years, and ex-FX GBV growth improved two points to 15% while the reported line slowed. None of that is currency. The mechanism is disclosed and repeatable: a product organization shipping 80% more features at 60% faster cycle time, with the host-pricing lever, universal single-fee migration and AI search all still ahead. Q3 nights are guided to low double digits, which would be the fastest of the coverage period.

Bear view: Q2 2025 was the weakest nights quarter of the prior year at +7%, so the comparison flattered the print, and the same three points of currency tailwind that flattered Q1 flatters the Q3 guide. Reported GBV growth decelerated from 19.2% to 15.7%. The product features that supplied roughly 3 points of nights growth still lap in the second half, and management's proposed replacement, host pricing tools, carries no number, no timeline and no rollout share.

Our take: The bull has the better of this one and we would say so at a lower price. The ex-FX GBV acceleration is the single hardest number to explain away, and core markets accelerating simultaneously across four countries is not a comp effect. The lapping risk we have carried since February looks materially smaller than it did, because the improvement has broadened well beyond the three features that started it. Our FY26 revenue estimate goes to $14.1–14.2B and our FY27 to $15.7–16.1B. The problem is not the business.

Debate: What Do the Adjacencies Actually Cost?

Bull view: Hotels are growing three times as fast as homes, work in unconstrained as well as constrained markets, and convert roughly 35% of first-time hotel guests into home bookers. Services are almost entirely partnerships, so fulfilment costs sit with the partner and Airbnb is running lead generation at negligible marginal cost. Development cycles have compressed from nine months to six weeks. This is category expansion at close to zero incremental capital.

Bear view: The cost finally appeared this quarter and it is in the take rate, which went from a guided full-year lift in May to guided flat in August because of customer incentives on new businesses. Featured hotels carry a price match guarantee and up to 15% guest credit, and Airbnb is winning hotel supply partly on an extremely favourable commission. Five quarters in, not one of the three adjacencies has disclosed a revenue figure, and the hotel-to-home conversion metric quietly changed definition and population in the same quarter the stock re-rated.

Our take: Both are describing the same trade correctly and the bear is describing the part investors have not priced. Buying guests through fee concessions instead of through marketing spend is legitimate and probably efficient, but it is customer acquisition cost by another name, and routing it through the take rate makes it invisible in the expense lines where analysts look for it. On our arithmetic the FY26 revenue effect is on the order of $200M. We would price the adjacencies as a modest net positive to the funnel and a modest net negative to near-term monetization, which is roughly how management has now guided them.

Debate: Does a 20x Forward EBITDA Multiple Leave Anything for Shareholders?

Bull view: This is now a mid-teens revenue grower with expanding margins, a guided nights line accelerating to double digits, structural cost deflation from AI, an investment-grade balance sheet, roughly $9.7B of net cash, and a share count down about 10% since 2022. Three consecutive guide raises with no walk-backs have earned management the benefit of the doubt. Businesses with that combination do not trade at 15x, and the multiple compressed to 15x in the first place because the market did not believe the acceleration was real.

Bear view: Enterprise value is now roughly 20.6x our FY26 adjusted EBITDA estimate against 16.6x in May, while our estimates rose about 3%. That is 24% of multiple expansion against 3% of earnings revision, which is a re-rating, not an earnings story. The margin is guided to a floor, not to expansion, and the CFO explicitly said incremental margin gets reinvested. The take rate is guided flat and the entire supply base is about to be repriced. Trailing free cash flow of $4,827M against a $113B market capitalization is a 4.3% yield.

Our take: The bear has the arithmetic that decides the rating. Our base case of $188–198 uses 19 to 20x enterprise value to FY27 free cash flow of roughly $5.78B, which is a full point above what we paid in May and is generous rather than punitive given the guided margin floor. That gives roughly 8% upside against a bear case near $141, or about 21% downside. Nothing here is a criticism of the company. It is an observation that a 17.4% single session moved a stock from cheap-for-the-growth to fair-for-the-growth in seven hours, and we do not have an edge at fair.

Model Update & Valuation Framework

Valuation framework: At the $178.07 close and the June 2026 fully diluted share count of 634M, the equity is worth approximately $112.9B. Cash, short-term investments and restricted cash of $12,136M less $2,476M of net long-term debt gives $9,660M of net cash, for an enterprise value of roughly $103.2B. Against trailing twelve-month revenue of $13,159M that is 7.8x sales; against trailing adjusted EBITDA of $4,617M it is 22.4x; against trailing free cash flow of $4,827M it is 21.4x. On our FY26 estimates it is roughly 20.6x adjusted EBITDA and 19.9x free cash flow. The free cash flow yield on the equity is 4.3%.

What actually changed in the multiple: at the May print the same figures were 18.4x trailing and 16.6x forward adjusted EBITDA. Our FY26 adjusted EBITDA estimate has risen roughly 3% since then and our FY27 free cash flow estimate roughly 3%. The share price has risen 25.9%. Roughly seven-eighths of the move is multiple, not estimates. That is a legitimate re-rating for a business that has proved its growth algorithm, and it is also the whole of our downgrade: we were paid to own the disbelief, and the disbelief is gone.

Why the base case is 19–20x: we paid 17–18x in May for a low-to-mid-teens grower with an expanding margin. The growth guide is now at least mid teens with nights accelerating to double digits and a demonstrated three-quarter record of raises, which supports a further point to a point and a half. Applied to FY27 free cash flow of roughly $5.78B and adding net cash on 634M shares, 19x gives $188 and 20x gives $198. We stop there rather than pushing to 22x because the margin is guided to a floor rather than an expansion path, the CFO has said incremental margin gets reinvested, and the take rate is now guided flat with the entire supply base about to be repriced.

Risk-reward: base $193 (+8.4%), bull $236 (+32.5%), bear $141 (−20.8%). The base-to-bear ratio is roughly 0.4 to 1, against 1.4 to 1 in May and 2.1 to 1 in February. A probability weighting of 55% base, 25% bull and 20% bear gives an expected return of roughly 8.6%, which is a market return with single-name risk attached. That is the definition of a Hold under our framework.

Thesis Scorecard Post-Earnings

Overall: Thesis vindicated on the business and exhausted on the price. Of the five bear points carried into this print, two were challenged and moved toward Contained, including the one we called the central bear case in February. All four bull pillars are Confirmed and one recovered from At Risk. This is the best operating quarter of the coverage period by a wide margin. It is also the quarter in which the market agreed, in a single seven-hour session, and paid for it.

Action: Downgrading to Hold from Outperform. To be explicit about what this is and is not: it is not a call that the business has deteriorated, that the guide is at risk, or that the adjacencies will fail. It is an observation that a 17.4% session moved enterprise value to roughly 20.6x our FY26 adjusted EBITDA estimate from 16.6x, on estimates that rose about 3%, and that our base case now implies roughly 8% against a bear case implying roughly 21% downside. The February upgrade captured 46.7% from $121.35; we are not going to hold the rating through the point where the arithmetic stops supporting it. We would return to Outperform on a pullback toward roughly $155 with the guide intact, on a quantified contribution from the host-pricing lever, or on a first disclosed revenue figure for hotels that supports the growth ratios management has been quoting. We would move to Underperform only on evidence that the universal single-fee migration is disrupting host pricing or supply retention, which is the one genuine execution risk in the second half.