A major AI cost overrun has highlighted the rising expense of enterprise AI adoption, with one company reportedly spending $500 million in a single month on Anthropic’s Claude chatbot after failing to set usage limits.
According to an Axios report citing an AI consultant, the unnamed client allowed unrestricted employee access to Claude without proper controls or caps on usage. This led to excessive consumption of AI “tokens”—the basic units used to measure processing and output—resulting in a bill of roughly ₹4,770 crore in one month.
The scale of spending has raised concerns over unchecked enterprise AI deployment.
The report explains that AI pricing is usage-based, and employees exceeding token limits or continuously prompting the system can rapidly escalate costs. It remains unclear how many employees were involved in the overuse.
The incident comes as several major companies reassess AI spending amid rising operational costs. Microsoft has reportedly cancelled most Claude Code licences and is shifting toward internal AI tools by June 30.
Uber has also said it exhausted its annual AI budget within five months. Amazon, meanwhile, has shut down an internal AI leaderboard after employees overused AI tools to gain rankings, increasing infrastructure costs.
Industry experts warn that without strict governance, quotas, and AI literacy, companies risk inefficiency and runaway costs driven by excessive or low-value AI usage.
