Airbnb is shipping 80% more software than it was a year ago, and almost half of its customer service tickets are now handled by AI rather than by seasonal staff.
Most large consumer companies are running AI pilots. Chesky hired the executive who led Meta's AI group, put AI tutors in front of his whole leadership team, and started running the company with an agent trained on hundreds of gigabytes of his own material.
"Here's a way of saying it's an existential risk to everyone, including Airbnb, if we don't use it."
Chesky co-founded Airbnb and still runs it. He took it public after an emergency pandemic financing valued the company at $18 billion, priced the IPO around $48 billion and watched it trade at $100 billion on day one, and the core business now does about $100 billion a year in bookings.
I listened to the full segment so you can skip it.
Here are the 6 takeaways that matter.
👤 Guest: Brian Chesky, Chief Executive and co-founder of Airbnb
🎙️ Host: David Faber, CNBC anchor, who interviewed him at Goldman Sachs' technology conference
📰 Published: 10 September 2026 on CNBC's Squawk on the Street
🟣 Apple Podcasts | 🔗 Episode page | ⏱️ 10 min
Key Takeaways
Airbnb ships 80% more software than a year ago and AI now handles almost half its customer service tickets
Customer service used to mean training seasonal workers on hundreds of problem types
Chesky splits the industry into companies that make AI and companies that apply it, and says the value ends up with the appliers
His comparison is electricity: the money was made by the people who used it, not the people who generated it
AI compresses a ten-year expansion into a few years, which is why Airbnb is selling event tickets and looking at long-stay living
He thinks the word superintelligence is a mistake, because it assumes human capability stays still
A child learns a hot stove from one data point; the human brain runs on about 20 watts against tens of gigawatts of compute
70% of Airbnb's business sits in five countries and four of those five are still accelerating
For every person who stays in an Airbnb, he said about nine stay in a hotel
His advice to the OpenAI and Anthropic founders is to assume they are the underdog
If technology doubles annually for a decade, that is a thousand times, so nobody can know who wins
1. 80% More Software
Chesky gave two numbers for what AI has already changed inside the company. "Number one, we are shipping 80% more software than we were a year ago." On support, where he said Airbnb has one of the hardest customer service problems in the world with hundreds of distinct problem types: "Almost half our tickets are now handled by AI."
The support economics used to depend on seasonal hiring. He said the company historically had to train seasonal customer service workers, and that an AI can instead be trained on the entire context of a case.
He attributes the company's relative growth to the rebuild rather than to the market. "And so I think what you're seeing is we are now growing faster than our competitors because we are becoming a more AI native company."
The change started with a hire from Meta. Chesky said he brought in the executive who ran Meta's AI effort at the beginning of this year — "He was the leader of Meta AI, essentially the Llama models" — and asked him to make Airbnb an AI-native company as though it had been founded today with AI from the start.
Asked by Faber what that means in practice, Chesky described his own working habits first. He said he had been using only ChatGPT, was moved onto Claude Code, loaded hundreds of gigabytes of his own information into a corpus and now uses it as an agent: "I started really using it as an agent to help me run Airbnb." Every other executive was given AI tutors.
2. Make AI or Apply AI
His central distinction is between building models and deploying them. "So there are those who make AI and those who apply AI. And I think right now we're seeing most of the benefit going to people who make AI."
He expects that to invert, and reached for electricity as the precedent. "Just like electricity, there's a lot of money making electricity, but those who used it to apply it to consumer applications was where all the value went." Airbnb, he said, is not a frontier AI company — it does not make AI — but is probably one of the leaders in applying it.
The risk framing is symmetrical, which is what makes it an opportunity. Asked whether the next 18 months bring real use cases, Chesky said the risk of not using AI falls on everyone: "But it's a risk to us. It's a risk to everyone else. And if it's risk to everyone else, it's an opportunity for us businesses we had no business going into. We can now go into."
3. Verticals AI Unlocks
The first concrete new business is event tickets. "We're selling tickets on Airbnb now. We had no business selling tickets, but now we can spin up new businesses." He said the company is also looking at longer-term living on the platform.
He put a timetable on how much AI compresses expansion. Chesky said a move into hotels "would have been a ten year journey we now can do in a few years."
Faber pressed him on why any of this produces faster growth from here, given revenue, margins and the share price have all already moved. Chesky's answer was the IPO story and the maturity of the core business, which follows below.
4. Against Superintelligence
Faber put the case against AI optimism directly to him — job loss, dependence eroding people's ability to think, cyber risk, hacking, bioweapons, and the resulting opposition to data centers. "Brian, it's nice to hear you be so positive, but what gives you the confidence you're going to be right."
Chesky did not claim confidence. "I'm not confident I'm right. I'm not confident that it's going to be positive. But I think it can be. It's like nuclear power. It can either light up a city or destroy a city. AI is not inherently good or bad."
He objects to the industry's own vocabulary. "I don't think anyone in Silicon Valley should use the word superintelligence. Superintelligence sounds like an inorganic species that's smarter than us. That will replace us." Faber countered that OpenAI's chief scientist had published a piece that weekend called The Alien Mind, about the lack of alignment.
His alternative model is that people move too. Chesky said the incorrect mental model is that humans are stagnant while AI gets more intelligent and leaves them behind, and that the hopeful version is "a cognitive revolution in which we actually ride the intelligence" with people staying in the loop. "I think when people bet on superintelligence, they're also betting a little against the human brain and the human power and the human mind."
He argued human learning is still far more efficient than model training. "In many ways, we are smarter than AI." A child touches a stove once and knows it is hot, where a model needs hundreds or thousands of examples, and on energy: "Our brain is powered by what, 20w of energy? AI needs tens of gigawatts of compute." His conclusion was that "AI is a wave, and we can surf the wave rather than get swallowed underneath the water."
5. From $18B to $100B
The valuation history is the reason he rebuilt the company. "Pandemic hits, we do this emergency financing in $18 billion valuation, right? We hope to get back to $30 billion market cap when we go public. The night before, we priced around 48 billion. We go public that day at 100 billion." He said he had to grow into that valuation.
His own metaphor for the rebuild is architectural. "Airbnb was like a ranch house. We had one business and we built the company like a one story house. I realized we need to launch ten, 20 more businesses. You can't build ten floors in a ranch house. You got to rebuild the foundation." He said the last three or four years went into that foundation, and the company is now ready to launch new verticals the way Amazon added categories beyond books.
The core business is less mature than he assumed. "But here's the most surprising thing I've learned. Our core business, our mature business, is not mature. 70% of our businesses are in just five countries, right? Four of the five countries are still accelerating."
The size of the unaddressed market is his growth argument. He said Airbnb does about $100 billion a year in bookings in the core business, and that for every person who stays in an Airbnb about nine stay in a hotel, which he called completely greenfield.
Supply is coming to Airbnb because of what hotels charge. "Not only they're coming to us because they don't like the duopoly that exists. They want another channel and we have lower commissions. We have younger travelers."
6. Advice for Sam and Dario
Asked what he would tell the OpenAI and Anthropic chief executives before their IPOs, Chesky led with hubris. "And I think you're most at risk when you're on top of the world, because I think hubris always gets the best of us." He said he knows Sam Altman well and Dario Amodei less well, and was careful to say he was not implying either has the problem.
His practical instruction is to play as the underdog. "Be paranoid and always pretend like you're the underdog the moment like a sports team, the moment you think you're going to win and you're complacent, you lose." He said you go into the game knowing it is nil-nil.
The humility argument rests on compounding. "And I think what I would tell them is if technology, the rate of technology doubles every year for ten years, that's a thousand x. So maybe we're just one year into a thousand year journey." On that arithmetic, he said, nobody can know who wins in ten years, and there could be many more companies like the two of them.
Bonus Insights
Chesky laid out three horizons for the product: travel, then living, then ways to meet people. He said nearly a fifth of Airbnb's business is already people staying longer than 30 days, which is the bridge into the living horizon, and that meeting people is the hardest problem of the three because of how hard it has become to meet anyone at all. Travel, he said, is just the beginning.
Faber noted at the top of the exchange that he had enjoyed this conversation more than he might enjoy the ones about margins, and used the fact that Chesky is now an established figure rather than an up-and-comer to set up the IPO-advice question.
Chesky's bottom line is that the money in AI will end up with the companies that apply it rather than the ones that build it, and that Airbnb has spent four years rebuilding its foundation so it can add verticals at AI speed rather than at the ten-year pace the old company would have needed.
Products, Companies & Tools Mentioned
Airbnb (Shipping 80% more software than a year ago, with AI handling almost half of customer service tickets and about $100 billion a year in core bookings)
ChatGPT and Claude Code (What Chesky himself uses — he moved from ChatGPT alone to running an agent on a corpus of his own information)
Llama (Meta's model family, run by the executive Chesky hired at the start of the year to make Airbnb AI-native)
OpenAI and Anthropic (Both heading for public markets, and the subject of his advice about hubris and underdog mentality)
Amazon (His analogy for launching verticals — he asked how big it would have been as only a book business)
Books & Resources Mentioned
The Alien Mind (An essay Faber said OpenAI's chief scientist published that weekend, on the lack of alignment in current models)
If this was worth your time, send it to someone who follows the name.
Get the latest market chatter as it happens: