Uber keeps AI spending stable even as its use of artificial intelligence expands across engineering, customer service, and consumer features, CFO Balaji Krishnamurthy said after the company reported second-quarter 2026 earnings on Wednesday. The trick: cheaper models, smarter defaults, and giving employees direct visibility into what AI costs.
Krishnamurthy said in prepared remarks that the company cut its cost per token ‘over the past several months’ by ‘setting better defaults for different use cases,’ routing lower-priority tasks to lower-cost models, and ‘letting employees more clearly understand and manage their spend.’ The result, he said, was that ‘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.’
The discipline follows a period of runaway consumption. Earlier this year, Uber’s CTO Praveen Neppalli Naga went viral after disclosing that the company had spent its Claude Code budget for 2026 within just a few months of the year starting.
How Uber Keeps AI Spending Stable While Scaling Output
The cost controls are running alongside a strong set of financial results. Uber’s Q2 2026 earnings press release showed gross bookings of more than $58 billion, clearing the top end of its guided range of $56.25–$57.75 billion and marking the fourth consecutive quarter of above-20% growth on a constant-currency basis. Trips grew 18% year-over-year in the same period.
Profitability also strengthened. Uber’s Q2 2026 earnings supplemental data showed Adjusted EBITDA of $2.8 billion, up 33% year-over-year, with the margin rising to 4.9% of gross bookings from 4.5% in the prior-year period. Non-GAAP operating income was $2.1 billion, up 40% year-over-year. Trailing twelve-month free cash flow exceeded $10 billion for the first time in the company’s history, according to the earnings press release.
A $1.6 billion investment revaluation benefit supplemented GAAP earnings in the quarter, per a StockTitan summary of Uber’s Form 8-K filing. GAAP diluted EPS came in at $1.17.
Thousands of Small Wins, Not One Giant Hit
CEO Dara Khosrowshahi used the earnings call to push back against the idea that AI’s value will arrive in a single transformative product. ‘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.’
One measure of that incremental approach: when customers open the app to request a ride, Uber now shows destination suggestions powered by AI. Three-quarters of the time, the app correctly predicts the destination. Khosrowshahi also highlighted Cart Assistant, which builds a grocery shopping cart inside the Uber app from a recipe or list and adjusts it based on customer feedback. Users of Cart Assistant have cart sizes that are twice as large as those of users who do not use it.
By Q1 2026, more than 10% of Uber’s production-ready code was being driven autonomously by AI coding agents, according to Uber’s Q1 2026 prepared remarks. The company has also deployed what it calls ‘agentic pods’ to find AI use cases across functions from finance to HR.
On the consumer side, Uber Premium grew 40% year-over-year in Q2 2026. Only 20% of consumers currently use both the Rides and Eats platforms, but those who do grow 1.5x faster than users of a single platform, according to a BigGo Finance report on the Q2 2026 earnings call.
Autonomous vehicles remain a small fraction of overall volume. Uber facilitates hundreds of thousands of AV rides each week, Khosrowshahi said, but that represents less than 0.5% of the 300 million trips completed through the app in the same period. Unlike AI chatbots, he noted, physical deployment of AI through robotaxis requires real-world testing and government approvals. Uber has committed $10 billion to autonomous vehicle investment, according to the Q2 2026 earnings call transcript.
Customer service is where the cost savings are most visible today. Uber laid off about 10% of its customer service workforce last month, citing AI-driven efficiency gains. Krishnamurthy pointed to the function 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.’
Uber’s recently announced agreement to acquire Delivery Hero is expected to extend its reach to nearly 100 markets and roughly double the number of markets where the company can offer both mobility and delivery services, according to the earnings call transcript. How analysts model the margin impact of the $10 billion AV commitment against that expanding footprint will be a test of whether the cost-per-token discipline can hold at scale.