$500 Million Spent On AI In 30 Days: From Uber To Microsoft, How AI Is Destroying Annual Budgets

The AI boom inside corporate America is running into a problem nobody wants to talk about out loud. Companies handed employees unlimited access to powerful AI tools, told them to use AI for everything, and set no limits. Now the bills have arrived, and they are staggering. From a single unnamed company spending half a billion dollars on AI in one month, to Uber burning through its entire annual AI budget before summer, the enterprise AI experiment is producing financial chaos at a scale that is forcing hard questions about whether any of this spending is actually working. 

Half a Billion Dollars in One Month 

An AI consultant told Axios that one of their enterprise clients accidentally racked up a staggering $500 million bill in a single month on Anthropic’s Claude after failing to implement spending caps or usage controls for employees. To be clear, that is not a full year figure. That is 30 days. 

Employees of the enterprise were granted open access to Anthropic’s Claude platform without any spending limits or oversight. They used advanced AI coding agents and resource-heavy workflows, rapidly escalating costs. Agentic AI workflows, where models autonomously execute multiple tasks, made the situation worse. 

Nobody was checking. Nobody was capping. And by the time someone looked at the invoice, the damage was done. Finance departments are now auditing token usage, AI access is being restricted by role, and teams are being told to reuse outputs rather than repeatedly generate new prompts. These are basic controls that should have been in place from day one. 

Uber Ran Out of AI Budget by April 

Uber’s story is just as striking. The company deployed Claude Code to 5,000 engineers and watched monthly usage rates climb to between 84 and 95 per cent by April 2026. Per engineer API costs reached between $500 and $2,000 per month. 

The result was that Uber burned through its entire $3.4 billion 2026 AI budget in four months. Its CTO said the annual budget had already been exhausted before the year was half over. Reports also noted that Uber had been running internal leaderboards ranking engineers by how much AI they used, not by what they actually shipped or built. That kind of metric encourages usage for the sake of usage, not results. 

Uber’s COO admitted the costs were becoming harder to justify. That is a remarkably candid admission from a senior executive at one of the world’s most prominent tech companies. 

Microsoft Pulled the Plug on Claude Code 

Microsoft sharply reduced internal Claude Code licenses after usage costs began to climb. Some engineers were reportedly generating between $500 and $2,000 in monthly AI costs per person. Microsoft has since steered more teams toward GitHub Copilot and internal tools that offer tighter cost control. 

The cancellation deadline for affected teams is June 30, 2026, with engineers in the Experiences and Devices division, covering Windows, Microsoft 365, Outlook, Teams, and Surface, redirected to GitHub Copilot. Even for a company that has invested $13 billion in OpenAI and openly embraced AI across its products, the Claude Code bill was simply too high to ignore. 

The Real Cost Nobody Is Counting 

Here is the uncomfortable truth behind all of this. Companies are citing AI’s ability to automate jobs as a cause for layoffs. But the CEO of CloudBees told Axios that workforce cuts may simply be the only lever companies can pull to offset their AI bills. 

In other words, people are losing jobs not because AI replaced their work, but because the company needed to free up the budget to pay the AI bill. That is a fundamentally different story from the one being told publicly. The AI spending crisis is real, the ROI is murky, and the reckoning inside corporate boardrooms has quietly begun. 

Also Read: AI-Generated Videos On YouTube Will Soon Get Automatic Labels And Stronger Disclosure Tags