OpenAI and Cerebras sign multi-year deal for AI computing deployment

OpenAI and chipmaker Cerebras have signed a multi-year agreement to deploy up to 750 megawatts of computing capacity using Cerebras’ wafer-scale systems to serve OpenAI customers. The rollout is expected to begin in 2026 and will take place in multiple stages.

According to Reuters, the ChatGPT parent will purchase computing power over three years from the chipmaker as it seeks to meet rising demand for AI services and strengthen its position in the AI market. The deal is valued at more than $10 billion.

The deployment will focus on high-speed AI inference, which involves running trained models to generate responses. Cerebras said its systems can deliver responses up to 15 times faster than GPU-based systems for tasks such as coding agents and voice chat.

The partnership follows years of collaboration between the two firms. OpenAI and Cerebras were both founded in the mid-2010s and have worked together since at least 2017, sharing research and early technical work.

Sachin Katti, Compute Infrastructure at OpenAI, said, “OpenAI’s compute strategy is to build a resilient portfolio that matches the right systems to the right workloads. Cerebras adds a dedicated low-latency inference solution to our platform. That means faster responses, more natural interactions, and a stronger foundation to scale real-time AI to many more people.”

The agreement would significantly expand the reach of its wafer-scale processor technology starting in 2026. The deployment is expected to support hundreds of millions of users, with the potential to reach billions over time.

The agreement comes as AI developers race to secure large amounts of computing power to support growing use of generative AI tools, including chatbots and automated agents, which require fast and reliable inference at scale.