Executive summary. Altman framed OpenAI's 2026 direction around compute, saying the company is buying huge amounts of compute, vertically integrating, and pushing to build datacenters around the world. OpenAI also reported broader cloud flexibility, more than 40% revenue from enterprise, and APIs processing more than 15 billion tokens per minute.
Compute Is The Core Leadership Choice
Altman wrote that OpenAI is buying huge amounts of compute, vertically integrating, and pushing to build datacenters around the world because it believes in a future of universal prosperity. The cited evidence makes compute buildout the central operating choice in this OpenAI leadership profile. For enterprise practitioners, that turns leadership into an execution question: the organization is not only consuming cloud capacity, it is describing a more vertically integrated infrastructure path for the systems customers depend on. The responsible-AI question follows from the same claim, because a worldwide datacenter push changes the scale at which safety, reliability, procurement, and availability decisions have to be governed.
Cloud Dependence Is Being Rebalanced
OpenAI said its amended agreement with Microsoft keeps Microsoft as OpenAI's primary cloud partner, makes Microsoft's IP license non-exclusive, and lets OpenAI serve products across any cloud provider. That structure gives OpenAI a broader product-delivery lane while preserving Microsoft as the primary cloud partner. In leadership terms, the evidence points to optionality: OpenAI is keeping a central cloud relationship while also creating room to place products across different cloud environments. For security leaders, that raises practical review points around where products run, which cloud dependencies sit behind them, and how vendor accountability is evaluated when delivery is no longer confined to one cloud lane.
Enterprise Scale Raises The Bar
OpenAI said enterprise now makes up more than 40% of its revenue, while its APIs process more than 15 billion tokens per minute. Those figures make Altman's leadership question practical because enterprise revenue and API throughput both depend on the compute strategy described in his own infrastructure framing. They also connect the Microsoft agreement to customer reality: if OpenAI can serve products across any cloud provider, enterprise buyers will evaluate the company through availability, integration, and operating resilience rather than through model access alone. The supported profile is therefore not a personality story, but a governance story about compute ambition, cloud flexibility, and enterprise-scale usage converging in one operating model.
FAQ
What did Altman say OpenAI is doing about compute?
Altman wrote that OpenAI is buying huge amounts of compute, vertically integrating, and pushing to build datacenters around the world.
How did OpenAI's Microsoft agreement change?
OpenAI said Microsoft remains its primary cloud partner, Microsoft's IP license becomes non-exclusive, and OpenAI can serve products across any cloud provider.
What enterprise scale did OpenAI report?
OpenAI said enterprise now makes up more than 40% of revenue and its APIs process more than 15 billion tokens per minute.
Sources
OpenAI — openai.com (2026-04-26)
OpenAI — openai.com (2026-04-27)
OpenAI — openai.com (2026-04-08)
