How did Uber burn its AI coding budget?

Uber CTO Praveen Neppalli Naga said the company’s rapid growth in AI coding-tool usage has already maxed out its full-year AI budget only a few months into 2026.

The development underscores a practical tension for companies adopting AI-assisted software engineering: benefits can arrive quickly, but costs—especially usage-based costs for coding assistants and related tooling—can scale just as fast. Uber’s statement implies that demand for AI coding features surged internally to the point where spending forecasts didn’t keep pace with adoption.

Why it matters

• Budget planning becomes harder: If usage expands faster than predicted, teams can hit spending caps early.
• Operational adoption isn’t “free”: Even when AI tools boost developer productivity, finance teams may need tighter controls.
• Procurement pressure increases: Organizations may renegotiate vendor terms, shift to different models, or add rate limits and usage policies.

In Uber’s case, the immediate consequence is that the company has already reached its annual spending limit, suggesting either a pause/slowdown in new usage or tighter governance over who can run which tools and how often.

What to watch next

If more companies report similar “budget burn” moments, AI-coding procurement is likely to shift toward clearer cost models—such as per-seat pricing, predictable quotas, or stronger monitoring of real consumption rather than relying on initial pilots. The story also highlights that infrastructure and model access are becoming line items that can constrain product and engineering decisions as much as performance does.