The OpenAI employee who uses the most tokens uses 100 billion tokens a month.

Business Insider reported on Tuesday that Sam Altman (샘 알트먼), OpenAI's chief executive, disclosed this at a media event and said the world's biggest token user is not an OpenAI employee.

Altman said OpenAI's biggest token user 6.5 years ago used about 100,000 tokens a month, and that this level was likely the highest in the world at the time. He said the situation has changed markedly. Altman said the monthly level of 100,000 tokens has now become about the global per capita average, and that OpenAI's top internal user consumes 100 billion tokens a month.

OpenAI's internal record was not the world's top mark. Altman said he had identified a user outside OpenAI who uses more tokens than that, adding that he personally found it embarrassing. That means an external user surpassed the internal record despite OpenAI's organisational culture of actively encouraging the use of AI models.

Within OpenAI, token use itself has become part of the culture. The company is known to have a leaderboard showing token usage rankings among employees, and some employees have shared their usage on X, formerly Twitter. OpenAI's structure of selling tokens directly has also dovetailed with this expansion in usage.

Cases exceeding the 100 billion tokens a month Altman cited have also been disclosed. Peter Steinberger, a developer of OpenClio, 공개ed a screen showing he used 603 billion tokens over 30 days, and the New York Times reported that one OpenAI employee used 210 billion tokens in a week.

Across the industry, tensions are also rising between expanding AI usage and controlling costs. Amazon scrapped its token-usage leaderboard, and Uber reportedly set usage limits after its chief operating officer said it was becoming increasingly difficult to afford spending. Altman also appeared mindful of this climate, saying a meme has been circulating among corporate customers along the lines of, "The company spent its entire 2026 budget in the first quarter. Can you make it more efficient?"

OpenAI also no longer views the cost issue as a peripheral matter. Altman said the company is looking for ways to keep improving model performance while delivering "greater value at lower cost". He said the cost issue was barely mentioned as recently as early 2026, but AI costs have now become an important issue.

This shows a trend in which AI companies promote expanding usage as a growth indicator, while corporate customers have begun to weigh real cost efficiency more heavily. As competition over token usage continues, the next task for AI companies, including OpenAI, is not only to drive more use, but also how convincingly they can lower the cost.