Sam Altman Reveals OpenAI's Top Token Spender Burns 100 Billion Tokens a Month — And Still Isn't the World Leader
Artificial intelligence is no longer a novelty reserved for researchers and tech enthusiasts. It has become an industrial-scale resource, consumed in staggering quantities by companies, developers, and individual power users alike. Nowhere is that reality more vivid than inside OpenAI itself — where, according to CEO Sam Altman, the company's single heaviest token consumer goes through a mind-boggling 100 billion tokens every single month. And yet, as Altman himself admitted, that number is not even enough to top the global leaderboard.
What Sam Altman Actually Said
During an enterprise event hosted by OpenAI on a recent Tuesday, Sam Altman offered a rare window into the internal scale of AI consumption at the company he leads. Speaking candidly, Altman revealed that OpenAI's top internal token spender now burns through approximately 100 billion tokens per month. To put that figure in historical context, he noted that just six and a half years ago, the company's heaviest token user was consuming around 100,000 tokens per month — a number Altman described as likely the highest in the world at the time.
"Today, 6.5 years later, that is about the per capita average," Altman said, underscoring the exponential growth in AI consumption that has taken place in less than a decade. The comparison is staggering: what was once a world-record level of usage is now roughly what an average user consumes.
But the most striking part of Altman's remarks was his admission that someone outside OpenAI surpasses even that 100 billion token monthly figure. He called this fact a personal "embarrassment" — a candid acknowledgment that OpenAI's own employees, despite being among the heaviest AI users on the planet, are being out-consumed by an external entity whose identity Altman did not publicly reveal.
The Staggering Economics of Token Spending
To truly appreciate what 100 billion tokens a month represents, it helps to understand the economics of AI token usage. Tokens are the basic units that large language models use to process and generate text. A single token is roughly equivalent to four characters or about three-quarters of a word in English. At current API pricing for advanced models, 100 billion tokens could easily translate into hundreds of thousands — or even millions — of dollars in monthly compute costs, depending on the model being used.
This context makes earlier reports about OpenAI employees' spending habits even more telling. Peter Steinberger, the creator of a developer tool called OpenClaw, reportedly spent $1.3 million worth of tokens in a single month. That figure attracted widespread attention at the time, but it now appears to represent just one data point in a much larger pattern of extreme AI usage that has taken hold across the industry.
Token Budgeting: From a Side Note to a "Huge Issue"
One of the most practically significant points Altman made during the enterprise event was his acknowledgment that token budgeting has recently become a "huge issue" for companies deploying AI at scale. This is a notable shift in how enterprise AI adoption is being discussed publicly. For much of the past two years, conversations around AI deployment focused primarily on capabilities — what models could do, how accurate they were, and how quickly they could respond.
Now, cost management is increasingly entering the conversation as a first-order concern. As organizations integrate AI deeply into their workflows — using it for customer service, software development, data analysis, content generation, and more — the cumulative token consumption can escalate rapidly. A single automated pipeline running complex queries against a powerful language model can accumulate millions of tokens in a matter of hours.
Altman's framing of token budgeting as a "huge issue" signals that even OpenAI is aware that cost efficiency is now as important as raw capability for enterprise customers. This awareness is likely to accelerate the development of more token-efficient models, better caching mechanisms, and smarter routing systems that match workloads to appropriately priced models.
Why the "Unknown World Leader" Matters
The mystery surrounding the external entity that outspends OpenAI's own top user is more than just an interesting anecdote. It points to a broader shift in how AI is being consumed across industries. If a single customer — whether a large enterprise, a government agency, or a hyperscaler building AI-powered products — is processing more than 100 billion tokens per month, it suggests that AI has crossed into a phase of truly industrial-scale deployment.
This has significant implications for infrastructure, energy consumption, and competitive dynamics in the AI industry. Companies like OpenAI, Google DeepMind, Anthropic, and Meta are not just competing on model quality; they are competing on their ability to serve massive, continuous token demand at acceptable cost and latency. The customer who is spending above 100 billion tokens per month is not just a big spender — they are a strategic prize that every major AI provider wants to serve.
What This Means for Businesses Adopting AI
For business leaders and technology decision-makers, Altman's disclosure carries a clear message: AI adoption at scale is both more powerful and more resource-intensive than many organizations have fully internalized. The companies that will gain the most from AI are not necessarily those that use the most tokens, but those that use them most strategically.
Developing a clear token governance strategy — understanding which workflows genuinely benefit from the most capable and expensive models versus which can run efficiently on lighter, cheaper alternatives — is becoming a core competency for AI-forward organizations. Monitoring token usage in real time, setting budget thresholds, and regularly auditing AI pipelines for inefficiencies are practices that will separate disciplined AI operators from those who find themselves surprised by runaway costs.
The Broader Picture: AI Consumption Is Accelerating
Altman's remarks at the enterprise event are a useful reminder that the AI industry is still in a phase of rapid, compounding growth. The jump from 100,000 tokens per month to 100 billion over six and a half years is not linear — it is exponential, and there is little reason to believe that trajectory will flatten in the near term. As models become more capable and more deeply embedded in daily workflows, token consumption will continue to climb.
For OpenAI, the fact that an external customer has surpassed even the company's own heaviest internal user is, in a sense, a sign of success. It means the platform has been adopted deeply enough that at least one customer is running workloads that rival those of the people who built the system. Whether Altman chooses to publicly name that customer in the future remains to be seen — but the benchmark they have set is one that the rest of the enterprise AI world will be watching closely.