OpenAI has announced a partnership with Broadcom Inc. to design and manufacture its first proprietary artificial intelligence (AI) processors in a $10 billion deal that marks a strategic shift as the ChatGPT maker seeks to secure critical computing resources amid explosive demand for its services.
Under the agreement announced Monday, OpenAI will design the chips and Broadcom will handle development and deployment with production scheduled to start in the second half of 2026. The companies plan to roll out 10 gigawatts worth of custom processors — consuming power equivalent to more than 8 million U.S. households or five times the Hoover Dam’s electricity output — with full deployment targeted for completion by the end of 2029.
The deal, first reported by Reuters last year, represents the latest in a series of major chip investments by OpenAI as the AI industry races to build increasingly powerful systems. Last week, the company unveiled a 6-gigawatt chip supply agreement with AMD Inc. that includes an equity investment option, days after disclosing that NVIDIA Corp. plans to invest up to $100 billion in the startup while providing at least 10 gigawatts of data-center capacity.
“Partnering with Broadcom is a critical step in building the infrastructure needed to unlock AI’s potential,” OpenAI CEO Sam Altman said in a statement.
Financial terms of Monday’s agreement were not disclosed. OpenAI did not specify how it would fund the deal.
The partnership places OpenAI alongside tech giants Alphabet Inc.’s Google and Amazon.com Inc., which have developed custom chips to meet surging AI demand while reducing reliance on NVIDIA’s expensive and supply-constrained processors. However, the strategy carries risks: Similar efforts by Microsoft Corp. and Meta Platforms Inc. have reportedly encountered delays or failed to match NVIDIA’s chip performance. What is more, analysts said custom processors don’t threaten NVIDIA’s near-term market dominance.
Still, the custom chip boom has transformed Broadcom into a major beneficiary of the generative AI revolution. The company’s stock has surged nearly six-fold since late 2022, and in September it announced a $10 billion custom AI chip order from an unnamed customer that market observers speculated was OpenAI.
Altman has frequently cited limited GPU availability as a bottleneck to the company’s development. In a post on X earlier this year, Altman said ChatGPT-4.5, which he characterized as OpenAI’s closest attempt at human-like reasoning, experienced deployment delays because of hardware shortages.
“I’m pretty sure y’all will use every one we can rack up,” Altman wrote at the time, outlining plans to add tens of thousands of GPUs in the near term and eventually scale to hundreds of thousands.
