Brian Chesky is no longer treating artificial intelligence as a feature layer for Airbnb. He is reportedly exploring a dedicated AI lab, and that says something important about where consumer internet founders think the next advantage will come from.
Airbnb has spent the past few years talking about AI in practical terms: customer support, coding assistance, search, discovery and better tools for hosts. Now Chesky appears to be looking one layer deeper. According to Bloomberg, the Airbnb co-founder and CEO is planning an AI venture focused on developing models, with an emphasis that could include user interaction and design.
That matters because Chesky is not just another executive adding AI language to an investor presentation. He runs one of the few consumer platforms with deep identity data, payments infrastructure, marketplace behavior, reviews and real-world intent. If he is willing to back a lab outside Airbnb's ordinary product machinery, it suggests the next phase of AI may be less about bolting chat windows onto existing apps and more about rethinking how people make decisions online.
The Information has also reported that Chesky is in early talks to fund a new AI lab and is expected to remain CEO of Airbnb. That distinction is important. This does not yet look like a departure from Airbnb or a clean break into a standalone OpenAI-style company. It looks more like a founder creating optionality at the edge of his main business, where the research can be ambitious without immediately being buried under quarterly product deadlines.
Chesky has been unusually direct about the limits of today's AI products for travel and commerce. On Airbnb's first-quarter 2026 earnings call, he said the company did not believe anyone had figured out AI for travel or e-commerce yet. His criticism was not that the models are useless. It was that chatbots are often the wrong shape for the job.
Travel is visual, comparative and social. People want to see homes, neighborhoods, maps, prices, cancellation rules and the tradeoffs between one place and another. A long text thread is a poor substitute for that. Booking a trip also often involves multiple people, while most AI assistants still behave as if one person is making one clean decision in isolation.
This is where Chesky's design background becomes relevant. Airbnb was not built by owning the underlying lodging inventory. It was built by making a messy market feel usable, trustworthy and human enough for strangers to sleep in each other's homes. If AI changes the interface of commerce, the company that understands trust, identity and intent may have a better starting point than a company that only understands model benchmarks.
Airbnb has already been putting AI to work internally. The company said AI tools wrote about 60% of the code its engineers produced in the first quarter of 2026, and Chesky has pointed to those tools as a way to build software for API partners faster than the company previously could. Its customer support AI has also been handling a meaningful share of issues in the U.S. and Canada, reducing the need for some human escalation.
Those are useful gains, but they are not the real prize. Customer service automation lowers costs. AI coding improves throughput. The more valuable question is whether Airbnb can create a product experience that understands why a guest is traveling, what kind of stay would actually work and which host can deliver it reliably. That is harder than answering a support ticket. It is also much harder for a general AI agent to copy.
Founders are building around the hyperscalers
Chesky's move also fits a wider pattern. Prominent founders are no longer content to wait for OpenAI, Google, Anthropic, Meta or Microsoft to define every layer of the AI stack. Some will still use outside models. Some will rely on open-source systems. But the important companies are beginning to ask where their own defensible research capacity should sit.
Airbnb has shown this instinct before. In 2023, it acquired GamePlanner.AI for just under $200 million, a deal that gave the company more AI talent at a moment when Chesky was saying generative AI could reshape the platform into something closer to a travel concierge. Airbnb also added Google's James Manyika to its board in 2023, another signal that Chesky wanted AI expertise close to the company's strategic center.
A dedicated lab would raise the stakes. Talent is the obvious constraint. Senior AI researchers have choices, and many are already inside heavily funded labs with enormous compute budgets. Chesky's pitch would have to be different. Airbnb cannot outspend every hyperscaler, but it can offer a real consumer problem, a founder-led culture and a chance to design AI for decisions that end in real-world experiences rather than screen time.
There is risk in that. AI labs can become expensive status symbols if they are not connected to product discipline. Airbnb has shareholders, a marketplace to run and regulatory issues in cities around the world. Chesky's strength since the pandemic has been focus, stripping the company back and making it more founder-led. The test is whether a new lab deepens that focus or distracts from it.
The market implication is straightforward. AI advantage is moving from access to models toward ownership of context, interface and trust. If Chesky can turn Airbnb's marketplace knowledge into a better AI travel experience, the company gains more than efficiency. It gains a stronger defense against the next generation of search, agents and booking assistants. That is what investors and competitors should watch next.
Also read: OpenAI gives ChatGPT a memory system that can keep preferences current • SK Hynix moves closer to a Wall Street listing • OpenAI is making ChatGPT remember users more like a regular assistant
