Uber Burns Its 2026 AI Budget In Four Months On Claude Code
Uber Burns Its 2026 AI Budget In Four Months On Claude Code
Publish Date: 2026-05-17 23:18:00
Source Domain: www.forbes.com
-
Unplanned Budget Overrun: Uber’s 2026 AI budget was exhausted by April, four months into the year due to the rapid adoption of Anthropic’s Claude Code by its engineers. The adoption rate far exceeded financial predictions.
-
Adoption and Usage: By spring, around 95 percent of Uber engineers were using AI tools monthly, with 11 percent of backend updates completed automatically by agents with no human intervention.
-
Non-Traditional Pricing Model: Unlike per-seat pricing used by other AI providers (like Microsoft), Anthropic’s token-based consumption model allows for unpredictable and high usage costs, as it largely depends on how heavily engineers use the tool.
-
Operational and Governance Gaps: Uber lacked formal AI governance policies and usage controls, which exacerbated the budget overrun. Only a fifth of the organizations have mature governance frameworks for AI tools.
-
Changing Pricing Models Across Industry: Following Uber’s experience, there’s an industry trend towards separating agent tools and third-party harnesses from subscription plans, introducing flat rate costs, and indicating that the previously low-cost usage is likely to rise.
-
Management and Accountability Issues: Organizational incentives like internal leaderboards for token usage further accelerated budget depletion because the teams driving adoption weren’t the same ones managing the costs.
-
Impact on CFOs: CFOs need to reconsider traditional budgeting for consumption-priced tools, recognize how incentive structures drive usage, and acknowledge that the predictive power of pilot programs does not extend to broader organizational rollout.
-
Future AI Adoption Challenges: The broader challenge for CFOs and organizations will be ensuring financial visibility and control over the unpredictable cost increases resulting from the growing use of agentic coding tools.
