Uber Let AI Write the Code. It Blew the Budget (aimmediahouse.com)

Uber's Chief Technology Officer, Praveen Neppalli Naga, revealed that the company underestimated the financial impact of using AI coding tools, such as Claude Code. Within months, Uber’s spending on AI coding surpassed its entire annual budget allocation. This shift reflects broader trends in the tech industry where AI tools are becoming integral to software development; around 95% of Uber's engineers are using AI tools regularly, and nearly 70% of production code is now AI-generated. With AI systems now responsible for an increasing percentage of code contributions, the dynamics of software engineering are shifting towards more supervisory roles for engineers who oversee the outputs of AI. The rapid integration of AI in Uber's development processes has led the company to reassess its budgeting, as costs are linked to usage rather than headcount, complicating financial planning. With AI coding tools operating under usage-based pricing models, Uber is now facing constraints as compute demands rise alongside more complex applications, such as rider-driver matching and dynamic pricing systems. As AI agents take on a larger role in code generation, companies like Uber are adapting their engineering workflows, which raises important considerations about cost efficiency and the evaluation of AI-generated code before deployment.