Xinhua News Agency, San Francisco, October 13 (Reporter Wu Xiaoling) The American Open Artificial Intelligence Research Center (OpenAI) and American semiconductor company Broadcom Corporation announced on the 13th that they have reached a strategic cooperation to jointly develop a customized AI accelerator with 10 gigawatts of computing power.

OpenAI announced on its official website that the company will be responsible for AI accelerator and system architecture design, and will cooperate with Broadcom to develop and deploy it. The project is scheduled to start in the second half of 2026 and is expected to be completed by the end of 2029. The two sides have reached a long-term agreement on the joint development and supply of AI accelerators.

According to reports, the system architecture will use Broadcom's Ethernet and other high-performance connection solutions to achieve vertical and horizontal expansion of AI data centers. It will be deployed in OpenAI's own facilities and partner data centers to meet global artificial intelligence computing power. Rapidly growing demand.

OpenAI believes that this collaboration highlights the importance of custom accelerators and Ethernet in building a scalable network architecture for AI data centers. Through self-developed chips, OpenAI can directly integrate its AI model and product development experience into hardware design, thereby improving overall performance and intelligence levels.

OpenAI CEO Sam Altman said this collaboration is a key step in building the necessary infrastructure to unlock the potential of AI and bring tangible benefits to individuals and businesses. Developing its own accelerator will further improve the partner ecosystem and jointly build the capabilities needed to promote the cutting-edge development of AI.

Charlie Kavas, president of Broadcom Semiconductor Solutions Group, said that the combination of customized AI accelerators and Ethernet solutions will optimize the performance and cost of the next generation AI infrastructure. (End)