OpenAI, the artificial intelligence research organization, announced on Monday a strategic partnership with semiconductor manufacturer Broadcom for the development and deployment of custom AI accelerator hardware. This collaboration is set to provision 10 gigawatts worth of specialized AI accelerator racks, a significant move to bolster its computational infrastructure.
Deployment of this new hardware is scheduled to commence in 2026 and will continue through 2029. The custom accelerators are slated for integration into OpenAI's proprietary data centers and those operated by its partners, aiming to optimize performance for its AI models. In a statement, OpenAI indicated that the rationale behind designing its own chips and systems is to "embed what it’s learned from developing frontier models and products directly into the hardware, unlocking new levels of capability and intelligence." This approach aims to tailor computing power more precisely to the specific demands of cutting-edge AI model training and inference.
Although specific financial terms of the agreement were not publicly disclosed by either party, the Financial Times has estimated the potential cost to OpenAI could range from $350 billion to $500 billion over the specified period. This initiative represents a substantial capital expenditure in advanced computing infrastructure, aligning with the industry-wide trend of increasing investment in AI hardware.
The Broadcom announcement is the latest in a series of major infrastructure deals recently secured by OpenAI. Last week, the organization confirmed an agreement to purchase an additional six gigawatts of chips from Advanced Micro Devices (AMD) in a transaction reported to be worth tens of billions of dollars. In September, Nvidia announced plans to invest $100 billion into OpenAI, which included a letter of intent for OpenAI to access 10 gigawatts of Nvidia hardware capacity.
Furthermore, reports in September indicated that OpenAI had entered into a $300 billion cloud infrastructure agreement with Oracle; however, neither company has publicly confirmed the details of this specific transaction. These multiple, large-scale partnerships underscore OpenAI's aggressive strategy to scale its hardware and cloud computing resources, reflecting the escalating demands for computational power in the development and deployment of advanced artificial intelligence systems across various industrial sectors.
