OpenAI Launches Compute Capacity Plan with Long-Term Agreements Lockdown

Artificial intelligence company OpenAI said on Wednesday that it will provide long-term computing power access assurance to its enterprise customers, aiming to meet the growing infrastructure needs of these businesses when constructing critical AI workflows and products on its platform.

The plan, named "Guaranteed Capacity," allows customers to sign one to three-year term commitment agreements in exchange for discounted token pricing and priority access to computing power across model product portfolios. The longer the annual commitment period, the greater the discount.

The Chief Executive, Sam Altman, said that as enterprises' reliance on AI systems deepens, customers are seeking more stability assurances from the company. "Customers are increasingly demanding certainty from us in terms of computing power," said Altman, "As model capabilities continue to improve, we expect global computing power supply to be in a tight state for a considerable period of time."

This plan is OpenAI's latest move to lock in large enterprise customers against intensifying competition from Anthropic, Google DeepMind, and AI services supported by Amazon. The multi-year commitment agreements provide OpenAI with a more predictable revenue stream, while also securing supply and discounted pricing for enterprise customers during peak demand periods.

Under the plan, customers can flexibly allocate guaranteed computing power to cloud service providers and different model series supported by OpenAI, covering applications for production systems, user-facing applications, and AI agents – which have become the core areas for large enterprises to deploy generative AI.

The announcement comes as OpenAI continues to expand its infrastructure. Computing power supply is becoming a strategic constraint for the entire industry, with major cloud computing giants and AI labs competing to secure electricity, land, and semiconductor supplies. Computing power assurance pricing agreements help OpenAI more accurately predict server and chip procurement needs while providing the infrastructure certainty required for enterprise customers to build multi-year AI strategies.