OpenAI has introduced a 'Guaranteed Capacity' programme for corporate customers that allows them to secure computing infrastructure in advance for up to 3 years.

On May 26 local time, IT outlet TechRadar reported that the programme targets enterprise demand to secure stable computing resources needed for large-scale artificial intelligence services through long-term contracts.

The programme focuses on companies running large workloads that need uninterrupted computing rather than smaller businesses seeking to avoid intermittent disruptions. Eligible companies can request capacity exceeding 1 billion tokens per minute. Pricing varies depending on whether the contract term is 1, 2 or 3 years and on annual spending.

OpenAI explained that customers can use their contracted capacity across its product portfolio. The structure is designed to lock in long-term demand across multiple products rather than a single service.

The move comes as enterprise AI demand rises rapidly and pressure from intermittent service disruptions grows. OpenAI recently disclosed that about 40 percent of its $2 billion in monthly revenue comes from corporate customers. The company sees that share continuing to rise. Weekly ChatGPT users reach 900 million, but there have also been earlier Wall Street Journal reports that OpenAI failed to meet some internal targets for 2025.

OpenAI Chief Executive Sam Altman (샘 알트먼) said the programme reflects customer feedback and will be offered only until the currently allocated capacity is used up. That has been interpreted as meaning OpenAI will take long-term contracts within the limits of the infrastructure it has secured.

Altman acknowledged that large enterprise contracts and applications for guaranteed capacity also help the company's financial planning. He said predictable revenue would allow it to expand data centres more systematically and secure more consistent cash. He added, "We intend to maintain our intention to build as many computing resources as quickly as possible."

The trend is also tied to OpenAI's strategy to expand infrastructure. Microsoft and Oracle, among others, are participating in its flagship project, 'Stargate'. OpenAI unveiled the project in January 2025 and set a goal of securing 10 gigawatts of AI infrastructure in the United States by 2029, which it achieved in just over a year. In April, it said it had additionally secured 3 GW of capacity.

OpenAI and its partners plan to significantly increase compute capacity over the coming years. To that end, they are reviewing potential data centre sites across the United States. The company also highlighted sustainability improvements related to cooling facilities at its Abilene, Texas, campus.

Ultimately, the programme is structured to provide enterprise customers with a means of securing long-term compute resources while giving OpenAI predictable cash flow. It is seen as a strategy to lock in large AI demand through advance contracts while pushing ahead with data centre expansion and improvements in operating efficiency.

Introducing OpenAI Guaranteed Capacity: a new offering that enables customers to guarantee long-term access to OpenAI compute. We’ve made long-term investments in infrastructure, partnerships, and capacity planning to help customers scale reliably. Now, Guaranteed Capacity…