An unnamed large company has reportedly spent nearly $500 million on Anthropic’s Claude AI platform in a single month because it failed to set usage limits and spending limits for employees.
According to the report, granting unlimited access to the Claude AI tool within the company led to a sharp increase in token usage. As a result of employees’ extensive use of AI, costs have spiraled out of control and reached record highs.
Engineers working with complex agent-based workflows, large-scale queries, and parallel programming processes have incurred hundreds, and in some cases thousands, of dollars in costs each month. When this model is applied to thousands of employees, the total cost quickly grows to hundreds of millions of dollars.
According to experts, autonomous agents and extended thinking functions require more computing resources than regular chat usage. This leads to a significant increase in token consumption and, as a result, costs.
The incident is considered indicative of a broader problem in corporate AI use. It is reported that a number of large technology companies have also begun to reconsider AI projects due to high usage costs.
According to the report, Anthropic offers tools such as an admin panel, user limits and compliance controls for enterprises. However, these features require prior activation and it is believed that such control mechanisms have not been established in the company in question.
Experts believe that this incident will lead companies to implement stricter governance policies in AI projects. Many organizations are already working on spending limits, role-based access rights and real-time monitoring systems.
Experts emphasize that although AI increases productivity, it can pose a serious risk to company budgets if there are no necessary control mechanisms.
