OpenAI and Broadcom to Deploy 10 GW of Custom AI Accelerators

OpenAI and Broadcom Inc. announced a multi-year collaboration to co-develop and deploy 10 gigawatts of OpenAI-designed AI accelerators, announced in a press release. The partnership will deliver accelerator and networking systems for next-generation AI clusters, with deployment beginning in the second half of 2026 and completion expected by the end of 2029.

The systems will integrate Broadcom’s Ethernet, PCIe, and optical connectivity technologies for scale-up and scale-out networking. OpenAI will design the accelerators and systems, embedding its AI model expertise directly into the hardware. Broadcom will manufacture and deploy the racks across OpenAI’s facilities and partner data centers.

According to the companies, the collaboration builds on existing co-development and supply agreements and marks a focus on scalable and energy-efficient AI infrastructure. It follows OpenAI’s earlier $10 billion investment in developing its own AI chips with Broadcom, extending that effort into full-scale deployment.

Executives from both companies said the initiative will enable large-scale AI computing capacity, with Broadcom providing the end-to-end networking solutions and OpenAI focusing on custom chip design and system integration.