Group 1 - OpenAI and Broadcom have announced a long-term strategic partnership to develop and deploy custom AI accelerators with a total capacity of 10 gigawatts (GW) [1][2] - OpenAI will lead the design of the AI accelerators and system architecture, leveraging its expertise in cutting-edge AI model development to optimize chip performance for core scenarios like inference [2] - Broadcom will collaborate on the development and deployment, providing networking systems including Ethernet solutions to ensure the accelerator racks can meet large-scale computing demands through vertical and horizontal scaling [2] Group 2 - The initial deployment of AI accelerators and networking system racks is planned for the second half of 2026, with the goal of completing the full 10GW capacity by the end of 2029 [2] - These racks will be deployed in OpenAI's own data centers and partner facilities, providing foundational support for global AI applications [2]
OpenAI与博通达成战略合作,2026年启动10GW级AI加速器部署