Uber CTO: Tokenmaxxing Era Ending Amid AI Cost Solutions

Uber's Chief Technology Officer, Praveen Neppalli Naga, has indicated that the company has found a sustainable approach to managing its artificial intelligence expenditures, suggesting a shift away from the "tokenmaxxing" trend. This trend, characterized by companies incentivizing extensive AI tool usage, led Uber to exhaust its entire 2026 AI budget within the first few months of the year. Naga admitted in an interview with The Information that he had to "go back to the drawing board" on spending after the rideshare company actively encouraged employees to utilize its AI tools, notably Anthropic's Claude Code. This encouragement included the implementation of "leader boards" to rank software engineers based on their AI usage, a practice that contributed to the rapid depletion of funds.

Despite the initial overspending, Naga stated that Uber has now developed strategies to deploy AI more efficiently without incurring prohibitive costs. He shared on X (formerly Twitter) that the company is observing "very interesting trends on AI costs," which he believes signals the "end of the so-called ‘tokenmaxxing’ era." Uber achieved this by quadrupling the number of employees using frontier AI tools, which paradoxically lowered the cost per token. Key to this cost reduction were improvements in prompt caching, adjustments to default model settings, the evaluation of new, more efficient models, and providing engineers with visibility into their AI usage and associated hourly costs. Naga emphasized that cost increases were not a consequence of restricting access but rather by treating efficiency as an engineering challenge rather than solely a budgetary one.

The increasing pressure for companies to demonstrate a return on their substantial AI investments is becoming more pronounced. This comes as the stakes for AI's return on investment (ROI) are rising. For instance, Jim Reid, global head of macro and thematic research at Deutsche Bank Research Institute, issued a warning last month that significant AI productivity gains are still several years away. The financial implications are substantial, with profit margins for the "Magnificent Seven" companies experiencing a notable increase, swelling from 15% to 25% between the first quarters of 2023 and 2026. This period highlights the growing financial expectations tied to AI advancements and the need for companies to manage these costs effectively. Uber's approach, focusing on engineering solutions for AI cost management, offers a potential model for other organizations navigating similar challenges in the rapidly evolving AI landscape. The company's success in reducing per-token costs while increasing adoption suggests a more mature phase of AI integration is emerging, where efficiency and strategic deployment are prioritized over unchecked usage.