OpenClaw creator spent $10 million on AI tokens last month | InfoWorld
Pete Steinberger, the creator of the OpenClaw AI digital assistant, reportedly spent a staggering $10 million on AI tokens last month. This significant expenditure was disclosed during a conversation with Cloudflare CEO Dane Knecht at the Cloudflare Connect 2026 Conference in San Francisco. Steinberger, now employed by OpenAI, incurred these costs while utilizing OpenAI's token-based services, highlighting the substantial resources involved in advanced AI development.
A key detail from the discussion was Steinberger's admission that "not all of that is useful," suggesting a degree of experimentation or inefficiency inherent in the development process. He also raised a pertinent question regarding infrastructure, musing, "Why is this still using my computer? Shouldn’t this all be in the cloud?" This indicates a desire for more scalable, cloud-native compute environments for AI tasks, moving beyond local machine limitations and potentially hinting at future architectural shifts for agentic workloads.
For the OpenClaw ecosystem, this revelation underscores the immense operational costs associated with developing and iterating on sophisticated AI agents. The expenditure by a single prominent developer signals that scaling multi-agent systems or complex, long-running agentic workflows could lead to prohibitive expenses for many organizations. It also emphasizes the critical need for frameworks and methodologies that prioritize token efficiency and cost optimization within agentic AI development.
This signal demands attention from a broad audience. Developers building agentic AI systems should scrutinize their token usage and explore optimization strategies, while researchers must focus on creating more efficient models and architectures. Operators and CFOs, in particular, need to understand these real-world costs to accurately budget for AI initiatives and evaluate the economic viability of deploying advanced agentic solutions at scale.