OpenAI is collaborating with Broadcom to develop and deploy 10 gigawatts of custom AI accelerators. OpenAI will design the chips and systems, leveraging insights from its AI model development. Broadcom will handle the development and deployment, which is slated to begin in the latter half of 2026 and conclude by the end of 2029.
This partnership aims to meet the increasing global demand for AI, with deployments planned across OpenAI's facilities and partner data centres. By designing its own chips, OpenAI seeks to optimise hardware for AI workloads, improve efficiency and reduce reliance on third-party GPUs. Broadcom will provide networking and hardware IP, with the new systems based on in-house accelerators paired with Broadcom's technology.
This agreement expands OpenAI's hardware partnerships, complementing existing collaborations with Nvidia and AMD. The move signifies a growing trend among major AI players to explore custom silicon solutions.
