Uber Exhausts AI Budget in Four Months Amid Rapid Claude Code Adoption

Uber's recent experience highlights the financial risks associated with the rapid adoption of advanced AI tools. By April, the company had completely exhausted its artificial intelligence budget for 2026, just four months into the year, after seeing a surge in the use of Anthropic's Claude Code among its engineering teams.

This significant budget overrun was confirmed by Chief Technology Officer Praveen Neppalli Naga, who noted that the speed at which Claude Code spread among approximately 5,000 engineers far surpassed the company’s financial projections. In 2025, Uber's total research and development expenditure reached $3.4 billion, a 9% increase from the previous year. However, the budget collapse was not due to scaling issues but resulted from a pricing model that enterprise finance teams struggled to manage effectively.

The Challenge of Consumption-Based Pricing

At the heart of Uber's budgetary crisis is the consumption-based pricing model linked to Claude Code. Unlike traditional software pricing, typically based on fixed licenses, the costs of Claude Code depend on the volume of tokens consumed during operations. This leads to significant variability in expenses, complicating accurate cost predictions for finance teams. For example, engineers using the tool for simple tasks like autocomplete may incur minimal costs, while those managing complex workflows can face expenses ranging from $150 to $2,000 per session.

The rapid uptake of Claude Code at Uber was remarkable. When it was rolled out in December 2025, only 32% of engineers were utilizing it by February; that figure surged to 84% by March, with 95% of engineers using AI tools monthly. Notably, around 70% of committed code was generated through these tools, and an astonishing 11% of live backend updates were produced entirely by AI agents, without any human involvement.

Cultural dynamics further exacerbated the financial miscalculation. Engineers were ranked on internal leaderboards based on their usage of Claude Code, incentivizing higher token consumption and accelerating budget depletion. The disconnect between the teams driving software adoption and those managing the associated costs proved to be a critical flaw in Uber's strategy.

Implications for AI Budget Management

Uber's situation underscores broader implications for organizations adopting AI technologies. Many enterprises lack formal governance policies for AI tools, with only 43% having established guidelines. Even fewer — just 21% — possess mature governance structures for agentic workflows. The absence of strict spending controls, such as budget caps and real-time monitoring of token usage, contributed to Uber's rapid budget burn when deploying Claude Code on a large scale.

CFOs across industries should take note of Uber's experience. Pilot programs often do not accurately reflect the costs incurred when scaling AI usage, especially with consumption-priced tools. Incentive structures, such as leaderboards that promote usage without limits, can lead to significant financial liabilities. As Anthropic prepares to implement a credit-based billing system on June 15, this indicates a shift in the industry towards more predictable pricing models.

Looking Forward: The Future of AI in Finance

Despite its budgetary challenges, Uber remains committed to its AI initiatives. Naga mentioned plans to explore OpenAI's Codex alongside Claude Code, envisioning a future where agent-driven engineering manages coding, testing, and deployment with human oversight. However, the pressing question for finance teams is whether they can effectively monitor and control costs associated with these AI tools as their use expands.

As the landscape of AI consumption pricing evolves, organizations will need to adjust their financial planning strategies. Procurement teams should pursue fixed-rate agreements rather than relying on consumption pricing to ensure predictability in budgeting. The lessons learned from Uber's experience will undoubtedly influence how enterprises manage AI costs in the future, as the demand for advanced AI solutions continues to rise.