Corporations Rein In AI Usage as Token Costs Surge Beyond Budgets

Major corporations are scaling back their use of artificial intelligence as the cost of running AI models climbs. Companies like Microsoft, Uber, Meta, and Amazon have started limiting AI usage after internal spending on tokens surged. The shift marks a departure from earlier enthusiasm when executives urged employees to maximize AI adoption.

Real-world costs challenge AI enthusiasm

Nvidia CEO Jensen Huang previously told engineers they should spend half their annual salary on AI tokens to stay productive. He compared avoiding AI to using paper and pencil for chip design. That mindset is now being challenged by real-world costs.

An anonymous consultant told Axios that one client accidentally spent $500 million in a single month on Claude usage because it failed to set spending limits. Uber operations chief Andrew Macdonald said it is 'getting harder to justify' AI spending. Microsoft canceled most of its Claude Code licenses.

Mixed productivity gains from AI adoption

Studies show mixed results from AI adoption. An OpenAI report found that AI improves productivity by about an hour a day. An MIT study of 350 public deployments indicates that 95% of them fail to turn a profit or achieve performance goals.

The trend is not limited to Big Tech. Internal AI token usage leaderboards at Meta and Amazon have been taken down, suggesting those companies are no longer encouraging heavy token consumption. The overall message is that AI, especially large language models, remains expensive and often unprofitable.

For now, the corporate AI push appears to be cooling. Companies are tightening budgets and demanding clearer returns on AI investments. The era of unrestrained token spending may be ending.