Uber CTO says surge in AI coding tools has maxed out full-year AI budget — just months into 2026

The short, sharp headline

It has been reported that Uber CTO Praveen Neppalli Naga warned the company’s rapid uptake of AI coding tools burned through its entire 2026 AI budget only a few months into the year. The comment — made in internal discussions and shared with reporters — allegedly shows how quickly expenses for large language models and code-generating services can balloon when teams lean in hard.

Budget blowout, and why it matters

Spending on code-focused LLMs isn’t pocket change. Developers calling on hosted models for heavy code generation, testing and iteration can rack up token and inference bills fast. It has been reported that Naga used examples — reportedly involving tools like Anthropic’s Claude — to illustrate how a single workflow can multiply costs. The result: a scramble to figure out whether the productivity gains justify the price tag. Cue the CFOs clutching their calculators.

Ripples across tech

This isn’t just an Uber problem. Firms across Silicon Valley are grappling with a new calculus: model quality versus recurring cloud and API fees. Do you standardize on cheaper, smaller models? Host on-prem? Tighten guardrails? The emotional moment here is unmistakable — excitement about AI’s promise colliding with the cold reality of invoices. The chatbot gold rush may be cooling into a cost-conscious era.

What to watch next

Will Uber renegotiate vendor terms, throttle usage, or invest in internal models to control spend? Those are the pragmatic responses to watch. One thing is clear: the honeymoon with pay-as-you-go LLMs is getting more complicated. Will engineering teams trade a little convenience for budget sanity? Time — and the next expense report — will tell.

Sources: theinformation.com