ChatGPT Employee Spends $10 Million On Tokens In 30 Days - Trak.in - Indian Business of Tech, Mobile & Startups
Peter Steinberger, who created the open-source AI assistant OpenClaw and now works on OpenAI’s technical team, has revealed that he used about $10 million worth of AI tokens in just one month. He shared this during a conversation with Cloudflare CTO Dane Knecht at the Cloudflare Connect 2026 conference in San Francisco.
This huge amount shows how expensive it can be to run advanced AI agents on a large scale. Steinberger also admitted that not all the work produced through such heavy token use is useful.
Who Is Peter Steinberger and What Is OpenClaw?
Peter Steinberger created OpenClaw, an open-source AI agent built to do tasks beyond answering questions. AI agents can use software, carry out a series of steps and automate tasks, with different levels of human supervision.
Steinberger joined OpenAI earlier in 2026. Work on OpenClaw was expected to continue through an open-source foundation.
His work includes testing ways to use AI in software development. AI agents can help write code, review other people's contributions, manage tasks and do other software engineering work.
Why Did He Spend $10 Million on AI Tokens?
AI companies usually charge customers based on tokens, which are small units of text or other information that AI models process. Costs can grow quickly when an AI system repeatedly examines large amounts of information, writes code, checks results or runs several agents.
During their conference discussion, Knecht pointed out that Steinberger had access to an almost unlimited budget for tokens. Steinberger replied that people often do not understand that much of the work produced this way may not be useful.
The reported $10 million is Steinberger's estimate of how much his AI token use cost during the previous month. It does not mean that a typical OpenClaw user would receive a bill of the same size. The report also does not give a detailed breakdown of the models used, the tasks performed or the number of tokens consumed.
AI Agents Could Move More Work to the Cloud
Steinberger also questioned why AI agents still depend so much on users' computers instead of running entirely in the cloud.
AI agents hosted in the cloud could work continuously, coordinate tasks and run without needing a user's computer to stay switched on. However, moving this work to the cloud also raises other concerns, including infrastructure costs, data privacy and security.
For businesses, the challenge is to decide whether the extra work completed with autonomous AI systems is worth the cost of running them.
Not Every AI-Generated Result Is Useful
The report about this spending also points to an important limit of AI automation: using AI more does not always create more value.
AI agents may repeat tasks, produce unnecessary content or make extra calls to AI models to check their own work. This can raise costs without leading to similar improvements in productivity.
Businesses using AI therefore need to track token use, set spending limits and check whether automated tasks are delivering results they can measure.
AI Agents Need Better Security Controls
Steinberger also discussed the dangers of giving AI agents too many permissions. An agent that can access files, apps or private information could cause security problems if it makes a wrong decision or follows harmful instructions.
He suggested that one AI agent could check the permissions of another. This could help decide whether an action is needed and whether it follows the rules set to keep things safe.
This approach could add another layer of supervision, but it would not remove the need for standard security measures and human responsibility.
What This Means for the Future of AI
The reported $10 million spent on tokens in one month shows both what AI agents can do and how expensive it can be to make them work more independently. Businesses are exploring ways to automate software development and other tasks, but the cost remains an important limit.
The next stage of AI adoption will depend not only on more capable models, but also on cheaper ways to run them, more efficient computer systems and better ways to make sure automated work provides real value.
Summary
Peter Steinberger, the creator of OpenClaw, estimated that he used $10 million worth of AI tokens in one month, according to his remarks at Cloudflare Connect 2026. He now works at OpenAI and discussed the costs of running AI agents, the need for cloud-based systems and the importance of security measures. He also admitted that not all work produced by AI is useful. This highlights the challenge of balancing AI's abilities, its cost and the productivity gains it delivers.