When AI costs begins to spiral: Company accidentally spends $500 million on AI in a month

A large company accidentally spent $500 million (₹4750 crore) in just one month on Anthropic's Claude AI because it did not set any limits on employee usage. The information was revealed by an AI consultant and reported by Axios on Thursday, sending internet into a frenzy.

The company reportedly allowed employees to use Claude AI without spending caps, usage restrictions, or monitoring tools. Workers used expensive AI features such as coding assistants, AI agents, and long text processing tasks.

As thousands of employees used these tools at the same time, costs increased rapidly and reached half a billion dollars within 30 days.The incident has raised concerns among business leaders about the growing costs of AI. 

Microsoft reportedly reduced many internal Claude Code licenses after AI costs became too high. Uber also said it spent its entire AI budget for 2026 by April because of heavy use of AI coding tools, as per reports.

At Amazon, the company recently shut down an internal AI leaderboard after employees began using AI unnecessarily to improve their rankings. The practice became known as tokenmaxxing (unnecessary use of AI), increasing costs without providing real business value.

Without proper planning, organisations may spend heavily on AI tools without improving productivity or profits. Several technology leaders have reportedly warned that AI investment must be linked to measurable business results.

Many companies are now introducing approval systems, usage tracking, and employee training programs to make sure AI tools are used responsibly and efficiently. These measures can help prevent unexpected expenses and improve the return on AI investments.