OpenAI CEO Sam Altman revealed that artificial intelligence costs have transformed from a non-concern into an urgent crisis for enterprises in just a few months, with major companies like Uber and Walmart now capping employee AI usage due to unexpectedly massive bills. During an enterprise event this week, Altman acknowledged the dramatic shift, noting that companies are approaching him with a now-familiar refrain: their entire 2026 budget was exhausted in the first quarter.

Why Did AI Costs Suddenly Become a Problem?

The speed of this transition has caught even Altman off guard. "The issue never came up. People were totally happy with the amount they were spending," Altman stated about the situation just months ago. "All of a sudden [AI costs] are a huge issue." The OpenAI CEO admits he is uncertain why the shift happened so rapidly, despite acknowledging its severity.

The most likely explanation lies in explosive growth in token consumption. Tokens are the basic unit of measurement for AI usage; the more work an employee does with an AI tool, the more tokens they consume. Companies typically purchase licenses with set token limits, and any overage results in additional charges. Altman revealed that six and a half years ago, OpenAI's heaviest user consumed about 100,000 tokens per month. Today, that figure represents the global per capita average.

The scale of current consumption is staggering. OpenAI's current top user consumes approximately 100 billion tokens per month, yet Altman noted this person is not even the global leader. He described the existence of someone outside OpenAI using even more tokens monthly as a personal "embarrassment." To put this in perspective, one company reportedly paid roughly $500 million on Claude AI usage in a single month.

How Are Companies Responding to Runaway AI Costs?

Major enterprises have begun implementing strict controls on AI spending:

• Usage Caps: Companies like Uber and Walmart have started limiting how much AI individual employees can use, a dramatic reversal from the aggressive AI adoption push of recent months.
• Leaderboard Shutdowns: Meta and Amazon, which previously maintained internal leaderboards tracking employee token consumption, have shut down these systems following mounting costs.
• Budget Exhaustion: Uber previously disclosed that its entire annual AI budget for 2026 was consumed within just a few months of the year.

Altman acknowledged that OpenAI is working to provide more value for its users and make token use more efficient. However, he emphasized that the cost crisis emerged unexpectedly, without clear warning signs until enterprises suddenly faced bills far exceeding their projections.

The timing of this cost crisis carries significant implications for OpenAI's planned initial public offering (IPO). Concerns about unsustainable spending patterns could become a liability as the company prepares to go public. Rival Anthropic has already filed for an IPO at a reported trillion-dollar valuation, adding competitive pressure to OpenAI's own public market ambitions.

Interestingly, OpenAI appears to maintain its own internal token leaderboard, even as other major tech companies have abandoned theirs. This suggests the company is tracking consumption patterns closely, possibly to understand and address the cost explosion that has caught the industry off guard.