OpenAI teams up with AMD in multibillion-dollar deal to expand AI compute capacity

OpenAI has entered into a major partnership with AMD to expand access to the computing power needed for its artificial intelligence models. The deal, announced on Monday, will see OpenAI deploy 6 gigawatts of GPUs, with the first phase using AMD’s Instinct MI450 chips expected to begin in the second half of 2026.

OpenAI President Greg Brockman explained the motivation behind the deal, citing the increasing global demand for AI infrastructure. “We need as much computing power as we can possibly get, and I think the whole industry right now is trying to really adapt to the amount of demand for AI services that we see coming over the next couple of years,” he told CNBC.

Brockman has previously stated that OpenAI hopes to eventually provide every user with access to their own GPU, a goal that would require an enormous expansion of compute resources.

“What we’re really seeing is a world where there’s going to be absolute compute scarcity, because there’s going to be so much demand for AI services and not just from OpenAI, really from the whole ecosystem,” he said. “And so that’s why it’s just so important for this whole industry to come together and say, how can we build in advance of this compute desert that we’re heading towards otherwise.”

The announcement follows repeated remarks from OpenAI executives this summer about the need for more computing power. CFO Sarah Friar described “constantly being under compute” as the company’s biggest challenge.

“This partnership is a major step in building the compute capacity needed to realize AI’s full potential,” said OpenAI CEO Sam Altman. “AMD’s leadership in high-performance chips will enable us to accelerate progress and bring the benefits of advanced AI to everyone faster.”

Brockman noted that while OpenAI continues its strong collaboration with Nvidia, working with AMD is key to building additional capacity. Nvidia recently announced plans to invest up to $100 billion in OpenAI.

“I think that Nvidia has something very special,” Brockman said. “We use them for training, we use them for inference. AMD, I think, also is really delivering in terms of the next generation of chip that they’ve been working on.”

As part of the new agreement, AMD has granted OpenAI a warrant for up to 160 million shares of AMD common stock, contingent on performance milestones. AMD CEO Lisa Su said the arrangement reflects the long-term nature of the partnership.

“The idea is that as OpenAI deploys compute, that’s a huge benefit to AMD in terms of the number, the amount of revenue that we get from that,” Su said. “Every gigawatt of compute is significant, double-digit billions of revenue to us.”

Following the announcement, AMD shares surged by more than 38 percent in early trading on Monday.