OpenAI has announced that it currently has no plans to deploy Google's tensor processing units (TPUs) for its products, despite early testing. This statement comes after reports indicated that OpenAI was considering using Google's AI chips to meet increasing demand.
OpenAI has partnered with Broadcom to develop and deploy 10 gigawatts of custom AI chips over the next four years, addressing the startup's significant computing requirements. The collaboration will utilize Broadcom's Ethernet technology and will be implemented in both OpenAI's and third-party data centers.