Peter Steinberger is on a token spending spree.
On Friday, the creator of OpenClaw posted a screenshot of his CodexBar, a tool that shows token spending on different AI coding tools. The image showed his token spending on OpenAI’s API at just under $20,000 for the day.
In 30 days, Steinberger had spent $1.3 million worth of tokens on OpenAI’s API, according to the image.
Steinberger notably doesn’t have to pay out of pocket for the tokens thanks to the fact that he now works at OpenAI. He wrote that the token funds were “perks of OpenAI supporting OpenClaw.” In response to an X user asking if OpenAI charged him for tokens, he responded “ofc not.”
But Steinberger’s AI bill is the latest example of just how costly intensive use of AI can be, and how access to free compute has quickly become a selling point in the AI talent wars.
High token bills are also all the rage as tokenmaxxing culture continues to sweep Silicon Valley. OpenAI is one of many companies to reportedly have a competitive token leaderboard.
Still, for many commenters on X, seeing over $1 million in AI spending for one engineer was shocking.
“Someone’s burning through enough tokens to bankroll a small startup,” one user commented. Others asked whether his token bill could be allocated to salaries for new engineers.
The majority of his spending is around the development of OpenClaw, Steinberger wrote, the viral AI agent sparked a Mac Mini buying craze and has since become the fastest-growing open-source product of all time. Steinberger also shared a screenshot of his GitHub, which showed dozens of projects from a device sleep tool to a system for AI agents to make phone calls.
Some social media users chalked Steinberger’s screenshot up to a form of marketing that could incentivize others to spend more. “You shipped nothing,” one user wrote. “You seem to have a very particular definition of nothing,” Steinberger wrote back.
He later posted again on X to acknowledge that people were “freaking out over my AI spend.” He wrote that he was trying to answer the question: “How would we build software in the future if tokens don’t matter?”
Steinberger then listed a variety of skills he spends tokens on. AI agents listen to his meetings and start working based on what he says. They also review comments for spam.
“All that automation allows us to run this project extremely lean,” Steinberger wrote.
In the comments, some users said that a project spending $1.3 million in tokens a month couldn’t be “lean.”
The number may actually be bigger, too. On his original post, one user asked whether the number pictured was actually his usage.
“Yup!” he replied. “At least on this account.”
