Uber announces end of tokenmaxxing era
Company's CTO shifts focus from AI adoption to efficiency in cost management.
1Tokenmaxxing era comes to a close
Uber's CTO Praveen Neppalli Naga declared the end of the company's tokenmaxxing period. This trend, which emerged in early 2026, encouraged employees to maximize their use of AI tools. Some corporations even made AI usage a performance metric. However, Uber will now prioritize efficient usage over sheer token volume, marking a significant shift in corporate AI strategy.
2Fourfold increase in AI adoption
According to Naga's data, the number of Uber employees using frontier AI tools has quadrupled since the beginning of the year. Despite this surge, the company managed to reduce per-token AI costs. This achievement resulted from strategies like improved prompt caching processes and giving engineers better visibility into their AI usage patterns.
3Cost reduction strategies
Uber implemented multiple approaches to lower AI expenses. These included using better default models, experimenting with open-weight models, and providing engineers with enhanced usage visibility. CFO Balaji Krishnamurthy reported during Q2 earnings that developers showed productivity gains, with engineer code output doubling compared to previous quarters.
4Industry-wide implications
The end of tokenmaxxing affects not just Uber but the entire tech sector. Companies like Coinbase are adopting model switching strategies, using frontier models for complex tasks while offloading simpler tasks to cheaper alternatives. This AI spending challenge has spawned new businesses focused on helping corporations reduce costs through various means.
5Uber's AI budget overrun
In April, Naga revealed the company had already exhausted its 2026 budget for Anthropic's Claude Code. He had previously announced in March that 1,800 weekly code changes were entirely written by Uber's internal coding agent. However, by May, COO Andrew Macdonald admitted the company was struggling to justify the ROI on its AI investments.
6Productivity-cost imbalance
Macdonald noted that increased AI expenditures weren't yielding proportional productivity gains. This challenge isn't unique to Uber; the entire tech industry is grappling with obtaining better returns from inflated AI spending. Many companies face similar difficulties in effectively integrating AI into their workflows while maintaining cost efficiency.
7Emergence of new business models
Uncontrolled growth in AI spending has given rise to businesses specializing in cost reduction solutions. Some offer consulting services to help executives manage AI budgets, while others develop products like inference infrastructure. These companies aim to help organizations scale their AI implementations cost-effectively in this new era of financial scrutiny.
8Future lies in efficiency
Uber's move signals the beginning of a new phase in corporate AI usage. Companies will now focus less on token volume and more on how efficiently those tokens are used. This shift represents a maturation of AI in the business world and could pave the way for more sustainable practices across the industry.
Original article: Business Insider — Uber, 'tokenmaxxing' dönemini sonlandırıyor