Airbnb shares surged 15% Friday after the corporate delivered one in all its strongest progress quarters in years and raised its full-year outlook — a turnaround CEO Brian Chesky is instantly crediting to synthetic intelligence.
In an unique interview with CNBC after the earnings report, Chesky mentioned Airbnb will spend “much more” on AI tokens this 12 months than it initially forecast as a result of the price of inference “pales as compared” to the income and productiveness the corporate is getting again.
Airbnb is chopping product-development time by roughly 60%, transport about 80% extra options 12 months over 12 months and conserving headcount roughly flat whilst AI spending rises sharply, Chesky mentioned.
“I believe now it is secure to say AI is the most effective factor to have occurred to Airbnb,” he mentioned. “I believe we’re turning into an AI-native firm, and I believe that’s most likely the primary rationalization for our outcomes.”
That conviction marks a pointy flip for Chesky. A 12 months in the past, Chesky mentioned the query inside the corporate was basically, “Is AI good for Airbnb or is AI dangerous for Airbnb?”
The shift is notable for a CEO who has lengthy approached expertise extra like a designer than a conventional engineer.
Chesky went to the Rhode Island Faculty of Design and studied industrial design earlier than co-founding Airbnb. Years later, he would change into shut with former Apple design chief Jony Ive and OpenAI CEO Sam Altman — and ultimately introduce the 2 males, serving to spark the connection that led to their work collectively on AI {hardware}.
Now, Chesky is making an attempt to carry that very same design-first sensibility to Airbnb’s personal AI overhaul.
The corporate employed Ahmad Al-Dahle, Meta‘s former head of generative AI and a frontrunner of its Llama work, as chief expertise officer in January. Chesky mentioned Airbnb had been “possibly center of the pack in AI” earlier than Al-Dahle arrived with a mandate to assist make the corporate “AI-native.”
There are already measurable indicators of that shift. Chesky mentioned AI helps Airbnb entice extra bookings, make it simpler for hosts to record and value houses, and decrease customer-service prices.
The corporate is piloting AI-powered search, utilizing AI to generate customized itemizing highlights and solutions for visitors, and serving to hosts create and value listings. In customer support, 45% of visitors who work together with Airbnb’s AI agent by no means want to talk with a human agent.
“It is actually throughout the board: Extra demand, extra provide, cheaper customer support,” Chesky mentioned.
Internally, Airbnb can be intently monitoring how workers use AI. Chesky mentioned the corporate seems at particular person token utilization as one measure of adoption, however considers {that a} comparatively crude metric and focuses extra closely on the output of groups.
“What we’re seeing is that throughout the board, groups are considerably extra productive,” Chesky mentioned. He mentioned these positive factors started with engineers and have since unfold to product administration, design, advertising and marketing, and artistic companies. “I’ve so underestimated the influence of AI.”
That productiveness is displaying up in Airbnb’s hiring plans as effectively. Headcount is roughly flat 12 months to this point, whilst spending on AI rises sharply, and Chesky mentioned buyers ought to count on income to develop “so much sooner” than staffing for the foreseeable future.
“Our philosophy has been not essentially to make use of AI to have fewer folks, however to make use of AI to get extra out of the folks,” Chesky mentioned. He added that income per worker ought to proceed to rise.
The economics are central to Chesky’s more and more bullish view of AI. Whereas many shopper corporations are nonetheless making an attempt to determine easy methods to generate sufficient income to justify inference prices, he argues Airbnb has an unusually favorable mannequin.
“One of many issues is lots of people really feel like they do not know easy methods to generate income on shopper with AI,” Chesky mentioned.
At Airbnb, he mentioned, inference prices pale as compared with “the amount of cash we make on each reserving” and the extra income generated by sooner product improvement.
“We’re going to spend much more on AI tokens this 12 months than we forecasted,” he mentioned. “However that is nice as a result of the ROI is there, and subsequently our income is way increased.”
Airbnb remains to be making an attempt to be selective about the place it spends these tokens and the corporate is utilizing greater than a dozen AI fashions internally, Chesky mentioned. These embody fashions like Anthropic‘s Claude Code and OpenAI’s Codex. But it surely limits entry to some slower and costlier fashions when their further functionality will not be mandatory for the duty.
Chesky is especially bullish on open-source fashions for consumer-facing merchandise and mentioned frontier fashions stay necessary for the toughest issues, however most shopper duties don’t require the costliest programs.
“Shoppers largely don’t want frontier fashions for many issues,” Chesky mentioned. “It is a matter of throttling the precise job for the precise instrument.”
He declined to call the open-source fashions Airbnb is deploying in shopper merchandise, saying the make-up of the corporate’s expertise stack is turning into competitively delicate.
The larger query is whether or not AI finally turns into greater than a instrument inside Airbnb — and as a substitute modifications how vacationers attain the platform within the first place. As AI brokers change into higher at assembling itineraries and performing on behalf of customers, corporations like OpenAI and Alphabet may theoretically change into the interface via which vacationers uncover and e book lodging.
However Chesky is skeptical that chatbots will change Airbnb because the transaction layer.
Journey is visible, troublesome to match via textual content alone and infrequently deliberate collaboratively, he mentioned, all areas the place the standard chatbot interface stays weak.
“I don’t imagine the chat interface is the precise interface for journey,” Chesky mentioned.
He expects chatbots to change into necessary for inspiration and itinerary constructing, however mentioned he doesn’t see them turning into main reserving platforms “within the coming future.”
For now, Chesky mentioned he’s centered on utilizing AI to increase Airbnb’s personal progress runway. He advised CNBC that first-time bookers are rising on the quickest tempo in 4 years and mentioned the U.S. enterprise accelerated from the primary quarter. Resorts, in the meantime, are rising 3 times sooner than the corporate’s conventional house listings as Airbnb expands past the home-rental market it constructed its title on.
“We aren’t an organization whose finest days had been within the 2010s,” Chesky mentioned. “We’re an organization the place the most effective days are in entrance of us.”
Airbnb inventory chart
