Microsoft cut its engineers off from AI because the bill got too big — why AI might not take your job after all

Everyone has been warning that AI is coming for your job, and that it’s just a matter of time. But that story glosses over how costly running AI can be — and those costs have the potential to slow or even stop what’s felt like an inevitable takeover.

Two developments shook things up this week. Microsoft (NASDAQ: MSFT) — the company that poured about $13 billion into OpenAI (1) and writes up to 30% of its own code using generative AI — reportedly told engineers (2) in a major division to stop using an AI coding tool because the bills got too big. And Uber’s (NYSE: UBER) chief technology officer said the company burned through its entire 2026 budget for Claude Code and Cursor in just four months, according to The Information (3).

Must Read

• Jeff Bezos backs a platform that lets anyone invest in rental homes for as little as $100 — 6 ways to build wealth like a landlord without actually being one

• Dave Ramsey warns nearly 50% of Americans are making 1 big Social Security mistake — here’s what it is and 3 simple steps to fix it ASAP

• Vanguard’s outlook on U.S. stocks is raising alarm bells for retirees. Here’s why and how to protect yourself

Sure enough, it sounds like the AI companies themselves are fully aware of the costs. Bryan Catanzaro, VP of Applied Deep Learning Research at Nvidia (NASDAQ: NVDA) — the $5+ trillion company making the chips powering much of the AI industry — told Axios (4) “for my team, the cost of compute is far beyond the costs of the employees.”

AI replacing human workers is still a real long-term risk. But here’s the thing: The companies actually deploying it at scale are openly admitting AI is too expensive, and that is an important signal.

What Microsoft actually did, and what it didn’t

Back in late 2025, Microsoft gave thousands of its people — engineers, product managers, designers, and even folks in non-technical roles — access to Claude Code (5), Anthropic’s command-line AI coding agent. The idea was to let them experiment and start coding with it. It spread pretty fast, way beyond just the technical teams.

Then the bills arrived.

Microsoft is now canceling Claude Code licenses (6) across its Experiences and Devices group — the team behind Windows, Microsoft 365, Outlook, Teams, and Surface — with a June 30 cutoff (7), the last day of Microsoft’s fiscal year. The company is moving its engineers to GitHub Copilot CLI (8), Microsoft’s more affordable in-house tool.

To be clear, this isn’t Microsoft taking a step back from AI. Hardly the case: Claude models still work inside Copilot CLI. And Microsoft’s broader deal with Anthropic (9) is untouched, including Microsoft’s up to $5 billion investment in Anthropic and Anthropic’s $30 billion commitment to buy Azure compute capacity. That deal stands, according to Fortune (2).

The problem now is the pricing model. The token-based pricing charges per output, and when engineers use an AI agent for hours on complex coding tasks, those tokens pile up fast.

Read More: 7 top habits of 'quietly wealthy' Americans. How many do you follow?

Uber: When adoption works too well

Uber’s situation makes this concrete. In April, Uber CTO Praveen Neppalli Naga told The Information (3) his company had burned through its entire 2026 AI coding budget in four months.

“I’m back to the drawing board,” Naga said, “because the budget I thought I would need is blown away already.”

And it wasn’t because Uber mismanaged funds. Like Microsoft, Uber deployed (10) Claude Code to its engineers in December 2025. By March, about 84% of Uber’s engineers had adopted Claude Code and were classified as agent coding users (11).

According to The Information (3), around 70% of code committed at Uber now originates with AI, and 11% live backend updates are shipped by an agent with no human in the loop. Individual engineers were spending between $500 and $2,000 per month. The irony is that this happened because the tool worked. Engineers found the AI genuinely useful and made it part of their daily workflow. The budget didn’t collapse because engineers were wasting tokens, but rather because they were actually leaning on the tool, something many bosses across Silicon Valley have been demanding of their employees.

Hype meets reality for AI economics

Catanzaro’s comment at Nvidia isn’t an isolated data point. Big Tech firms have collectively announced $740 billion in capital expenditures this year — that’s a 69% jump from 2025, according to Morgan Stanley (12). But Yale Budget Lab reports (13) there’s still no widespread data showing AI actually drives productivity gains at scale.

A 2024 MIT study (14) looked at the economics of automating vision-related work and found that AI could do it cheaply enough to make sense for about 23% of the wages tied to those tasks. For the remaining 77%, it was still cheaper to keep a human doing the work.

Keith Lee, an AI and finance professor at the Swiss Institute of Artificial Intelligence’s Gordon School of Business, told Fortune (4) what we’re seeing is “a short-term mismatch” driven by hardware and energy costs pushing up operating expenses for AI providers.

The infrastructure required to run AI at scale is projected to cost $5.2 trillion by 2030, according to McKinsey (15). “It’s not just about AI becoming cheaper than humans,” Lee said. “It’s about becoming both cheaper and more predictable at scale.”

What this means for workers worried about AI

None of this means AI displacement isn’t real. Tech layoffs are now more than 115,000 in 2026 so far across 152 companies, according to Layoffs.fyi (16), already on track to beat last year’s 120,000+. Companies are cutting jobs and investing in AI at the same time, even when the AI isn’t clearly saving money yet.

What Microsoft and Uber show us is a real constraint: to replace a human worker, AI has to deliver the same or better output for less money. Right now, for most jobs, that math isn’t quite mathing.

You May Also Like

• The tax breaks in Trump's 'big beautiful bill' expire after 2028 — and experts say most people won't act in time. What to do before the window closes

• 'Gold still crashing!': Robert Kiyosaki admits he was wrong — but doubles down on his $35K prediction

• Millionaires under 43 hold only 25% of their wealth in stocks. Here’s where their money is actually going

• I'm 49 years old and have nothing saved for retirement. What do I do? Don't panic. Here are 7 ways to catch up fast

Enjoyed this Moneywise story? Click Follow above for more. To get our best content and exclusive interviews first, join 250,000+ readers who subscribe to our weekly newsletter.

Article Sources

We rely only on vetted sources and credible third-party reporting. For details, see our ethics and guidelines.

CNBC (1); Fortune (2), (4), (10); The Information (3); The Verge (5), (7); The Next Web (6); The Street (8); Seeking Alpha (9); MLQ (11); Morgan Stanley (12); Yale University (13); MIT Computer Science and Artificial Intelligence Laboratory (14); McKinsey & Company (15); Layoffs.fyi (16)

This article provides information only and should not be construed as advice. It is provided without warranty of any kind.