Artificial intelligence is becoming more and more common in the corporate world, but if the usage is left unchecked, the costs can quickly add up. According to a recent report, a company invested almost half a billion dollars into AI solutions during one month because it decided to let its employees use Anthropic’s Claude unmetered. The employees used the AI tool to code and conduct research and other activities that require AI assistance. The other takeaway from the report was that employees were resorting to using complex AI workflows and long prompts which led to increased costs. This particular case has raised serious concerns among analysts about how businesses handle their adoption of AI technologies, as the associated costs could spiral out of control rather quickly once large amounts of employees start using it.
Axios recently reported that a major company allowed its workers to freely use Anthropic’s Claude for coding, research and automated task handling. As a result, the company reportedly spent nearly $500 million on AI tools in just one month. Moreover, when the employees were given a free hand, they relied on the advanced AI workflows that consume large amounts of computing power. Some teams also used long prompts containing huge volumes of text, which further increased costs.
Also read: Google tweaks Gemini usage limits after complaints: Here is what changed
Reacting to the incident, experts pointed out such cases may increase in the future, as when thousands of workers use AI tools at the same time without restrictions, the usage cost of the AI will rise rapidly. They also said that the real-time alerts and budget controls were either missing or poorly managed in this case.
Previously there were also reports that Microsoft cut back on the internal access to Claude Code after expenses reportedly climbed sharply for engineers using the tool regularly. Uber executives have also raised concerns about AI spending after large investments in AI coding systems reportedly exhausted internal budgets earlier than expected.
Other than that, Amazon has also shut down an internal leaderboard that tracked employee AI usage after workers allegedly began running unnecessary AI tasks to improve rankings. The trend has been informally labelled ‘tokenmaxxing’, where employees maximise AI usage even when the work offers little value.
Also read: Microsoft may unveil new coding focused AI model next week: Here is what we know
Industry experts are currently pointing out the trend that sees companies adopt AI without any specific objectives set in place. There is even a reported instance where certain staff members use enterprise-level AI technology to check the weather or compose elementary messages, something that could have been easily accomplished without an AI solution.
Executives and other industry insiders believe that enterprises are gradually shifting toward the adoption of measures such as expenditure alerts, limited access to expensive AI solutions and usage dashboards.