Why Uber Has Burned Through Its AI Budget (aimagazine.com)

Uber has exhausted its AI budget for 2026 due to unexpected costs driven by the extensive use of coding tools like Claude Code, which have become integral to the company’s engineering processes. The company's CTO, Praveen Neppalli Naga, reported that R&D expenses rose to $3.4 billion in 2025, with AI usage soaring as a cost factor. Engineers are now generating up to 11% of backend code updates solely through AI, leading to operational shifts and a reconsideration of the company’s financial strategies regarding AI investments. This situation highlights a broader trend in the tech industry, where firms are increasingly relying on AI for coding and development tasks—a practice known as “tokenmaxxing.” Uber's substantial monthly API costs, ranging from $500 to $2,000 per engineer, raise concerns about the sustainability of such AI tools. As AI systems begin to dominate the coding lifecycle, with projections indicating a future where “agent engineers” manage the entire development pipeline, Uber faces challenges in balancing innovation with financial viability. CEO Dara Khosrowshahi’s fears about AI replacing up to 80% of human roles underscore the long-term implications of this technological shift, particularly for millions of drivers currently using the platform.