Uber keeps AI spending stable using cheaper models and employees tracking cost, CFO says
AI is reshaping Uber — gradually, and as it keeps a handle on AI spending.
The ride-hailing company kept a lid on its token spending during the second quarter by “by setting better defaults for different use cases,” using lower-cost models for some tasks, and “letting employees more clearly understand and manage their spend,” CFO Balaji Krishnamurthy said in prepared remarks after the company reported quarterly earnings on Wednesday.
“As a result, cost per token has declined over the past several months, even as adoption has continued to increase, allowing us to keep overall AI spend broadly stable,” Krishnamurthy said.
Earlier this year, CTO Praveen Neppalli Naga went viral after saying that Uber had spent its Claude Code budget for 2026 after just a few months.
These days, Uber is also using AI to make lots of tiny changes to its app and offerings instead of taking big swings, CEO Dara Khosrowshahi said on the company’s earnings call on Wednesday.
One example: Using AI, Uber provides destination suggestions when customers open the app to request a ride. Three-quarters of the time, the app correctly predicts where the ride is going with those suggestions, Khosrowshahi said.
“A lot of people expect these technologies that are revolutionary — and AI is revolutionary — to have some giant hit,” he said.
Instead, “it’s going to be thousands of small hits and improvements to our ecosystem that’s going to drive, we think, growth for the foreseeable future,” Khosrowshahi added.
Uber is also reshaping its internal operations with AI. The company laid off about 10% of its customer service workers last month, citing the opportunity for efficiency gains from AI.
On Wednesday’s call, Krishnamurthy pointed to customer service as “a clear area where we should be able to up the quality of our customer support interactions as well as reduce the cost of our effort.”
The company has also used “agentic pods” to find uses for AI from finance to HR.
Uber’s consumer-facing uses of AI are also progressing, though they are unlikely to come all at once, Khosrowshahi said on the call.
Uber facilitates “hundreds of thousands” of rides in self-driving cars each week — less than 0.5% of the 300 million trips that happen through the app in the same period, the CEO said.
By contrast, AI has already become common for many online search users, Khosrowshahi said. Unlike AI chatbots, Uber’s physical use of AI through robotaxis requires real-world testing and government approvals, he added.
Uber is also experimenting with other uses of AI that “improve the fundamental experience of the consumer,” Khosrowshahi said. He pointed to Cart Assistant, which can create a shopping cart of grocery items in the Uber app from a recipe or shopping list and edit the cart based on feedback from customers.
Cart Assistant users have cart sizes that are twice as large as those of users who didn’t use the AI tool, Khosrowshahi said.
“You should expect AI to contribute to average order size, the quality and reliability of our service as well, and putting the right product in front of you at the right time,” he said.
Do you have a story idea about Uber? Contact this reporter at abitter@businessinsider.com or via encrypted messaging app Signal at 808-854-4501. Use a personal email address, a nonwork WiFi network, and a nonwork device; here’s our guide to sharing information securely.