Airbnb's AI Rebuild Is Fast. But Where Is It

I've booked well over a hundred Airbnbs. I've used the platform in places where the alternative to a clean, bookable room was genuinely uncertain — a guesthouse with no online presence, a guesthouse that maybe didn't exist anymore, a taxi driver's cousin's place that may or may not have had a second bed. Airbnb, at its best, functions as a kind of distributed trust infrastructure for travelers who don't want to be tethered to hotel zones. That's what it was built on. That's why people like me kept using it.

So when CEO Brian Chesky calls AI "the best thing to ever happen to Airbnb" — as he did this week, per CNBC's MacKenzie Sigalos — I'm paying attention. Not because I distrust him, but because I know this platform well enough to care where it's actually going.

The headline numbers from Airbnb's Q2 2026 earnings are hard to dismiss. According to BigGo Finance's recap of the earnings call, revenue surged 17% year-over-year to $3.6 billion, gross booking value climbed 16% to $27.2 billion, and nights and seats booked grew 10%. The company raised its full-year guidance. Shares hit a four-year high. By any financial measure, something is working.

Chesky's argument is that the something working is AI — not just deployed as a feature, but baked into how the company builds things. The example he keeps returning to is a telling one: it took Airbnb up to nine months to develop its groceries feature, but just six weeks to roll out airport pickups, according to Skift. I've used that airport pickup feature. The first time, I was landing in a city I didn't know, at an hour when the legitimate taxi queue had dissolved into a scrum of guys with unmarked sedans, and I genuinely wanted a name and a number already in my phone. The feature existing at all — quickly enough to be there when I needed it — mattered. Six weeks versus nine months isn't just a stat. It's the difference between a feature that catches a travel behavior while it's still relevant and one that ships into a market that's moved on.

TechCrunch reported that earlier this year, Airbnb disclosed that AI is now writing 60% of its code — though TechCrunch's piece doesn't specify whether that figure came from an earnings call or a specific executive statement, so I'm taking it as reported rather than primary-sourced. What's notable, though, is the gap TechCrunch identifies between the internal rebuild and the consumer-facing product: Airbnb is moving fast behind the walls, but the app you're actually using hasn't changed as dramatically. Not yet. The company is testing a new AI-powered search function, but it's still in controlled rollout. The engine has been rebuilt; the car looks pretty much the same from the outside.

Needham analysts noted that Airbnb has been rebuilding itself to move faster with AI, while Wedbush said the initiatives were speeding up product launch cadence and driving cost efficiencies, according to reporting from Lufkin Daily News and New Orleans CityBusiness. When two separate Wall Street shops with no particular incentive to agree reach the same conclusion about an operational story, that's worth registering — not as proof, but as a signal that Chesky's internal narrative is at least holding up to people who get paid to poke holes in it.

But here's the number that keeps pulling at me, and it has nothing to do with code generation speed.

According to Chesky's own comments, reported by CNBC, hotel bookings on Airbnb are now growing roughly three times faster than home rentals. Chesky surfaced this figure directly, so the attribution is his. Read that again: on the platform that was supposed to make hotels irrelevant, hotels are the fast-growing category.

For a travel writer, this is not an abstract data point. Airbnb was culturally built on a specific promise — not just cheap lodging, but different lodging. The apartment above a fish market in Lisbon. The converted stable in the Dolomites. The host who leaves you a note about which bar the locals actually go to. That promise was always imperfect, always uneven, sometimes flat wrong. But it was directionally distinct from booking a Marriott. The fact that hotels — the exact format Airbnb was supposed to disrupt — are now the platform's fastest-growing category tells you something real about where the user base is drifting, or being steered.

It could mean Airbnb's hotel inventory has simply gotten better and more competitive on price. It could mean travelers who tried the quirky apartment once have started craving the reliability of a hotel and are routing that booking through Airbnb out of loyalty or convenience. It could mean the "live like a local" framing always overstated how many people actually wanted that, and the platform is finally converging on what most of its users wanted all along: a room that works, priced well, searchable from the same app. Any of those is a plausible reading. What it doesn't look like is the original vision still running at full velocity.

Speed can obscure direction. AI giving Airbnb the ability to ship features in six weeks instead of nine months is genuinely useful — for the company, and potentially for the people using it. A faster iteration cycle means bad features fail faster, good features scale faster, and the platform can adapt to how travel is actually changing rather than how it was changing two years ago when the roadmap was written. That's real. I'm not dismissive of it.

But faster build cycles can't resolve the issues that have chased Airbnb for years — unpredictable cancellations, cleaning fees that turn a reasonable nightly rate into something eye-watering, the slow hollowing-out of neighborhoods where every apartment is a de facto hotel. Those aren't engineering problems. They're structural tensions between Airbnb's marketplace incentives and the communities it operates inside, and no amount of AI-generated code changes the underlying pressure.

What AI can do is make the features that paper over those tensions arrive faster. A better search function that helps you find a listing that's less likely to cancel. A smarter pricing layer that at least makes the fee structure legible before you've already invested twenty minutes in a booking. Faster dispute resolution. These are meaningful improvements, if they ship and if they work. Whether they shift the platform's character or just make the existing character run more smoothly is a different question — one that travelers, not engineers, will ultimately answer with their bookings.

Chesky calls this the best thing to ever happen to Airbnb. Maybe. The Q2 numbers give him reasonable grounds to say it. But faster shipping is genuinely good, assuming the destination is still worth reaching.

Kael Maddox is BuzzRAG's Adventure & Solo Travel Correspondent.