Uber Burned Through Its Entire 2026 AI Coding Budget by April

The counterpoint to the SaaStr miracle. Uber's full-year AI coding budget was exhausted by April. Four months into the year, the money was gone.

The math is straightforward and brutal. Uber saw 84% developer adoption of AI coding tools. At $500 to $2,000 per month per engineer, multiplied across thousands of engineers, costs exploded to 6x what they were in 2024. Their AI budget was built on pilot-stage economics. Pilot-stage economics don't survive contact with enterprise-scale adoption.

This isn't an Uber-specific problem. It's the canary. Every enterprise that built AI budgets on 2024 assumptions, when adoption was 15-20% and per-seat costs were lower, is heading toward the same wall. The productivity gains are real (Uber isn't cutting the tools), but the ROI models assumed costs would scale linearly. They didn't. They scaled faster than anyone modeled.

Uber is now hedging across multiple providers, splitting between Anthropic and OpenAI to avoid vendor lock-in and negotiate better rates. That's the enterprise playbook taking shape: multi-provider, usage-monitored, with hard spending caps that didn't exist six months ago.

Here's what connects this to the SaaStr story. SaaStr runs 20 agents at $257/month total and gets 140% revenue. Uber runs AI coding tools across thousands of engineers and blows the budget in four months. The difference isn't the technology. It's the scale. Small teams get asymmetric returns. Enterprise deployments get asymmetric costs.

If you're setting AI tool budgets for a team larger than 20, model at 3-5x your pilot costs. Not 1.5x. Not 2x. The adoption curve is steeper than you think, the per-seat costs are higher than you've been quoted, and your engineers will use the tools more aggressively than your pilot group did. Every CFO I've talked to who budgeted conservatively has been surprised. None of them pleasantly.

The industry needs honest cost data, and Uber just provided it, involuntarily.