OpenAI has partnered with Broadcom to co-develop custom AI hardware, with 10GW of chips set to be deployed by 2029.

In a statement, OpenAI said that the systems the duo are planning to develop will include AI accelerators and Ethernet solutions, with deployment targeted to start in the second half of 2026.

The news confirms reports from earlier this year that the two companies were working together on custom hardware design.

The solutions are expected to be fully rolled out by the end of 2029, with the capacity for the deployed custom AI accelerators totaling 10GW.

The financial terms of the deal have not been disclosed, but in early September, Broadcom’s CEO, Hock Tan, told analysts that the company had secured a fourth customer for its AI chip business that had committed to $10 billion in orders.

At the time, the FT reported that people with knowledge of the situation had confirmed the unnamed customer was OpenAI. However, on CNBC earlier today, Charlie Kawwas, president of Broadcom's Semiconductor Solutions Group, said OpenAI was not the unnamed customer.

In a statement announcing the partnership, OpenAI said that by designing its own chips and systems, the company can “embed what it’s learned from developing frontier models and products directly into the hardware, unlocking new levels of capability and intelligence.”

The racks incorporating the co-developed chips with be “scaled entirely with Ethernet and other connectivity solutions from Broadcom,” with deployments targeting OpenAI’s facilities and partner data centers.

“Partnering with Broadcom is a critical step in building the infrastructure needed to unlock AI’s potential and deliver real benefits for people and businesses,” said Sam Altman, co-founder and CEO of OpenAI. “Developing our own accelerators adds to the broader ecosystem of partners all building the capacity required to push the frontier of AI to provide benefits to all humanity.”

Charlie Kawwas, president of the semiconductor solutions group for Broadcom, added: “Our partnership with OpenAI continues to set new industry benchmarks for the design and deployment of open, scalable, and power-efficient AI clusters.

“Custom accelerators combine remarkably well with standards-based Ethernet scale-up and scale-out networking solutions to provide cost and performance-optimized next-generation AI infrastructure. The racks include Broadcom’s end-to-end portfolio of Ethernet, PCIe, and optical connectivity solutions, reaffirming our AI infrastructure portfolio leadership.”

This is the second major deal announced by OpenAI this month, following its agreement with AMD for the supply of hundreds of thousands of GPUs.

Under the terms of that deal – expected to kick off in the second half of 2026 – AMD will supply the generative AI giant with processors totaling 6GW of power consumption, with OpenAI saying it would build a 1GW data center using AMD MI450 chips, starting next year. However, no information has been provided about where this data center will be built or who the developer will be.

OpenAI has also been issued a warrant to purchase up to 160 million shares in AMD, structured to vest as specific milestones are achieved.

The month prior, OpenAI signed a letter of intent to deploy at least 10GW of Nvidia hardware in its data centers. Under the terms of that deal, Nvidia will invest up to $100 billion in OpenAI as the new systems are deployed. As with the agreement with AMD and Broadcom, the deal between Nvidia and OpenAI is also expected to commence in the second half of 2026.

OpenAI was valued at $500bn following an employee share sale in early October that saw current and former employees offload around $6.6bn worth of stock. Despite this, the company has yet to turn a profit, making it unclear how it plans to finance many of its recently announced deals or realize the $200bn of anticipated revenue it is forecasting for 2030.