OpenAI CEO Sam Altman stated that the company's highest monthly token usage user has reached approximately 100 billion. This figure represents a significant increase compared to six and a half years ago, demonstrating that the cost of model usage is rapidly becoming a real pressure as enterprises massively adopt generative AI.

Internal monthly consumption rises to 100 billion

At a corporate event on Tuesday, Altman said that six and a half years ago, OpenAI's internal "token consumption champion" used about 100,000 tokens per month. At the time, this was likely among the highest levels globally. Now, he says that this figure is roughly close to the global average usage, while the company's highest monthly usage has risen to 100 billion tokens.

External users spend more

Altman also stated that OpenAI later discovered that the number of tokens used by people outside the company even exceeded the internal record. He called this a "personal embarrassment," as OpenAI employees were known for their intensive use of their own models.

It has been mentioned multiple times before that OpenAI has a strong culture of model usage. Reports indicate that the company once had a token leaderboard, and employees would display their total usage on social media platforms. Some public cases suggest that internal consumption far exceeds 100 billion tokens.

• Peter Steinberger once spent approximately $1.3 million worth of tokens in a single month.
• Publicly available screenshots show that usage reached 603 billion in 30 days.
• The New York Times once reported that one employee spent $210 billion in a week.

Businesses are starting to tighten their budgets

Altman said that the cost of AI suddenly became a prominent issue in 2026. At the beginning of the year, customers were “completely satisfied” with their spending levels, and there wasn’t much discussion about it; but now, cost has become a “huge issue.”

Unlike OpenAI, which encourages high-frequency model usage, some large tech companies have begun to control internal token spending. Reports indicate that Amazon shut down its internal token leaderboard, and Uber also set a cap on token usage after management deemed related spending increasingly difficult to justify.

Altman also stated that OpenAI is continuing to advance its model capabilities and is attempting to deliver more value with lower expenditures. As enterprises shift their focus from rapid access to return on investment, AI cost control is becoming a new focal point of competition within the industry.