Airbnb graduates to AI native: A look at the metrics, ROI approach

Airbnb said it has graduated to be an AI-native company and its second quarter results and metrics back up that assertion.

The company last year laid out its infrastructure overhaul and plans to leverage AI throughout its operations and customer experiences. In the second quarter, Airbnb delivered revenue growth of 17% to $3.6 billion with net income of $816 million, or $1.37 a share. The earnings results were 12 cents a share better than Wall Street estimates.

CEO Brian Chesky said the company has evolved to offer more guest services and is seeing strong growth among Gen Z, a demographic that's new to the platform. Airbnb showed strength among multiple categories.

"Most people naturally want to point to 1 product or 1 initiative to explain the strong quarter, but that is not what's happening here. This is a culmination of changes we've been making over the last several years. We've rebuilt Airbnb from the ground up to be an AI-native company," said Chesky. "AI is transforming how we execute and build products. Said simply, AI is the best thing to ever happen to Airbnb."

Airbnb is projecting third quarter revenue growth of 155 to 17%. The company has added hotel booking that has added customers. Those customers often return to book homes too.

"What we're also seeing is about 35% of people who come to Airbnb and book a hotel for the first time, come back and book a home. So hotels not only are bringing new guests to Airbnb, but those new guests are sometimes often choosing to book a home. One is making the other stronger. And that's the basic theory of Airbnb," said Chesky.

Chesky cited a bevy of metrics and efforts Airbnb uses to measure its AI progress.

• The company reduced the time from concept to launch by 60%.
• Features and improvements shipped increased by nearly 80%.
• AI is accelerating the company's Project Hawaii, an effort to build improvements that compound over time. "AI is accelerating this work across search, sign-up, checkout and payments," said Chesky.
• AI is generating listing highlights, reviews and pricing recommendations for hosts. "We're rolling out AI tools that help new hosts get started faster and better understand their pricing and earning opportunities," said Chesky.
• 45% of the issues started with Airbnb's AI assistant are resolved without a human agent.
• In the second quarter, second quarter customer support cost per booking declined about 16% from a year ago.
• Expansion into new categories is faster. Airbnb Services added groceries as a service in 8 months to 9 months and airport pickups took 6 weeks to develop.
• Airbnb Experiences has added 1,000 new experiences for its most in-demand categories. The category is a small part of Airbnb revenue, but is seen as a growth avenue.

Chesky said Airbnb is entering its next stage of expansion that will move the company from the travel category to living. AI will be an accelerant to Airbnb where it was seen as a threat a year ago. "The existential risk to everyone was AI. Is AI good for you? Is AI bad for you? I think the moment of truth happened this year," said Chesky.

Airbnb’s evolution

• 2024: Airbnb CEO: GenAI's impact on app experiences minimal so far
• 2025: Airbnb: With tech stack in place, expansion plans accelerate | Airbnb: A look at its AI strategy

Airbnb hired Ahamad Al-Dahle, who formerly led Meta's Llama models, as CTO. "He came in, and I think we went from a company that was a middle of the pack company for AI to a leader in AI, at least amongst companies that are not frontier labs or hyperscalers," said Chesky.

Being AI native now enables Airbnb to go into new categories and develop products and services to attach to its core business.

Chesky said AI features on Airbnb will ramp. AI search is being tested for a small percentage of traffic and then will expand. He added that it will take some time to retrain the customer who is used to a search box.

Airbnb didn't provide a long term outlook, but the big theme is expansion.

AI inference costs

Not surprisingly, Chesky was asked about AI costs. He said most AI revenue since the launch of ChatGPT has been focused on the enterprise. Consumer companies have struggled to find value. "A lot of companies have not anchored that how to make money on the consumer side. Why is this? Because the inference cost is not cheap, and there's huge capital expenditures," said Chesky.

He noted that Airbnb doesn't have to spend on GPUs and inference costs are low relative to the ROI. CFO Ellie Mertz noted that just the customer service savings and slower hiring have justified the AI spending.

"Our transactions are very high dollar transactions. And 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," said Chesky.

He added:

"What you're seeing is the cost of tokens develop products and the inference costs on the models pales in comparison to the incremental revenue we generate and the incremental output or throughput we're seeing. I'm sure we can always be a little more efficient, but I think we're really disciplined. We're not like so-called tokenmaxxing, which was kind of a vanity of trying to have people use as many tokens as possible.

Luckily, I have a great technology leader in Ahmad, and we've been really, rigorous and thoughtful about -- it's not about how many tokens do you use. It's about the throughput of your product and the quality of your product designs and how much you're shipping."