Microsoft is pulling back on Claude Code for its employees. Uber burned through its entire 2026 AI budget by April. And if you're wondering what this means for the AI revolution everyone keeps talking about, here's the reality check: nobody figured out how to budget for this stuff.

Last December, Microsoft opened access to Anthropic's Claude Code across thousands of employees including developers, project managers, and designers. The AI coding assistant became wildly popular over six months. Too popular, actually.

According to reporting from The Verge's Tom Warren, Microsoft is now planning to reduce Claude Code usage in favor of its own GitHub Copilot CLI. The stated reason is that Claude Code "undermined" Microsoft's own coding tool. The unstated reason? The bill got too big.

Over at Uber, the situation is even more dramatic. The company exhausted its entire 2026 AI budget in four months after Claude Code spread to roughly 5,000 engineers faster than finance teams anticipated. CTO Praveen Neppalli Naga confirmed to The Information that the company is "back to the drawing board" on its assumptions.

Let's put that in context. Uber's total R&D spend hit $3.4 billion in 2025, up 9% year-over-year. The AI budget collapse isn't about scale. It's about a pricing model that enterprise finance teams have no idea how to manage.

Here's the problem: Traditional software licensing is predictable. You pay per seat, per year, done. AI tools charge based on usage, and usage scales exponentially when developers realize the tool actually works. Microsoft CEO Satya Nadella revealed last year that the company writes up to 30% of its code using generative AI. If you're paying per token or per request, that percentage translates into costs nobody modeled correctly.

The economics are genuinely unclear. Are AI coding tools cheaper than human engineers? In some cases, maybe. But when your AI budget for four months exceeds what you planned for twelve, and you're Microsoft or Uber with effectively unlimited resources, that's a red flag for every other company watching from the sidelines.

The bigger question is what happens next. Microsoft has the luxury of redirecting employees to its own tool. Most companies don't have that option. They're evaluating third-party AI tools with pricing structures that change quarterly and usage patterns that finance teams can't forecast.

This isn't a story about AI failing to deliver value. By most accounts, these tools work. Engineers love them. Productivity is real. The problem is that the value is hard to measure and the costs are hard to predict, which is a terrible combination if you're the CFO trying to justify the spend.

If Microsoft is pulling back and Uber blew through its budget in a third of the year, smaller companies with tighter margins should be paying very close attention. The AI tooling gold rush is running headfirst into budget reality, and nobody's sure what the ROI actually looks like yet.

The bottom line: AI coding tools are powerful. They're also expensive in ways companies didn't anticipate. And until someone figures out how to make the economics predictable, expect more stories like this.