OpenAI and Broadcom partner on AI hardware

OpenAI has forged a significant partnership with Broadcom, a leading semiconductor manufacturer, to develop a remarkable 10 gigawatts of custom AI accelerator hardware. Announced on Monday, this collaboration aims to enhance OpenAI's data centers as well as those of its partners, with deployment scheduled to begin in 2026 and extend through 2029. The initiative is poised to revolutionize AI capabilities, as OpenAI plans to integrate insights from its cutting-edge model and product development directly into the hardware. This strategic move is anticipated to unlock unprecedented levels of intelligence and performance, according to a statement from the company. While the specifics of the financial arrangement have not been publicly revealed, the Financial Times has speculated that the deal could involve an investment ranging from $350 billion to $500 billion. This partnership is just one of several recent infrastructure agreements for OpenAI. Just last week, the organization announced the acquisition of an additional six gigawatts of chips from AMD, a transaction valued in the tens of billions. In a further demonstration of its expanding infrastructure, Nvidia disclosed a remarkable $100 billion investment in OpenAI, which also included a letter of intent for OpenAI to access 10 gigawatts of Nvidia hardware. Additionally, OpenAI reportedly entered into a historic $300 billion cloud infrastructure agreement with Oracle in September, although neither company has officially confirmed this arrangement. TechCrunch has reached out for further details from OpenAI regarding these developments.