OpenAI has partnered with chipmaker giant Broadcom to develop custom processors for artificial intelligence, the company announced Monday. The chips are expected to launch next year and will deliver up to 10 gigawatts of computing power, roughly equivalent to the energy consumed by a major city.
The move comes amid a series of major tech partnerships done by OpenAI seeks to take tighter control over its infrastructure as demand for AI services surges. CEO Sam Altman has recently inked deals with Nvidia, AMD, Oracle, Samsung, and SK Hynix, focused on expanding AI data centers and chip supply. Still, a few of the agreements, including the one with Broadcom, have not disclosed financial details.
By designing its own processors, OpenAI hopes to improve performance and efficiency by tailoring hardware to its own AI models. Broadcom will co-develop and manufacture the chips, which will be deployed in OpenAI’s data centers and partner facilities.
The company’s aggressive expansion has drawn concern from energy experts, with AI data centers placing increasing strain on electricity grids. Despite strong investor enthusiasm and rising valuations, OpenAI is the world’s most valuable privately held company. The AI sector has yet to show signs of profitability.
The Broadcom deal follows a broader trend in the tech industry, as major players seek to reduce dependency on general-purpose chips. On Monday, Broadcom shares rose nearly 10% following the announcement, echoing recent market gains seen by AMD and Oracle after similar partnerships with OpenAI.