OpenAI announced its Guaranteed Capacity offering on May 19, allowing customers to reserve long-term access to AI compute power for products, agents, and workflows [1, 2].
Customers can choose commitment periods of one, two, or three years. Longer commitments receive increasing discounts, providing financial incentives for extended contracts [1, 2].
The service guarantees compute access based on each customer's spending level. This helps ensure capacity availability amid expected industry-wide constraints [2].
OpenAI CEO Sam Altman said, "Customers are increasingly asking us for certainty on capacity. As models get better, we expect that the world will be capacity-constrained for some time." He added, "The new offering will help OpenAI plan ahead, which I hope will be a big win-win" [1].
OpenAI will reserve enough compute capacity for key products like ChatGPT and Codex alongside the Guaranteed Capacity commitments [1]. The company said it will offer the service until current allocations sell out, with plans to relaunch it later [1].
Privately valued at more than $850 billion, OpenAI aims for a potential initial public offering in 2026. It told investors it targets about $600 billion in total compute spending by 2030 [1].
The Guaranteed Capacity program could form a basis for OpenAI's future compute business model, linking customer commitments to capacity planning and revenue streams [1].
