OpenAI, which is all over the artificial intelligence (AI) landscape, has inked a strategic deal with AMD that is expected to bring the GPU supplier more than $100 billion over a five-year period. Moreover, this multibillion-dollar partnership will allow OpenAI to acquire a 10% stake in AMD.
OpenAI is committed to purchasing 6 gigawatts of AMD GPUs, starting with the supply of 1 gigawatt worth of MI450 GPUs in the second half of 2026. OpenAI will buy 6 gigawatts of AMD GPUs either directly from AMD or through its cloud partners.
AMD claims its MI450 series will outperform Nvidia’s comparable offerings such as Rubin CPX through hardware and software improvements. “AMD GPUs, such as the MI355X, at the right price and on the right workload, already show a perf/$ benefit over the Nvidia B200,” said SemiAnalysis analyst Jordan Nano.
It’s worth noting that AMD’s collaboration with OpenAI began with the supply of MI300X GPUs in 2023 and continued with the MI350X series. MI450 will be AMD’s sixth data center GPU with AI capabilities and fourth “real” AI GPU following the MI300X, MI325X, and MI355X, which are all currently shipping and available for rent at prominent neoclouds.
In fact, OpenAI had already been onstage at AMD’s Advancing AI event in June 2025, talking about the company’s use of AMD GPUs. That’s also when AMD Chair and CEO Lisa Su called OpenAI a “very early design partner” for AMD’s MI450 GPUs.
AMD’s upside
The arrangement between AMD and OpenAI is a marvel of creative financing. AMD chief Su called it pretty innovative and added, “I wouldn’t say it came lightly.”
In this “multi-year, multi-generation” partnership, AMD will issue OpenAI a warrant to purchase as many as 160 million shares at an exercise price of $0.01 over time, provided that OpenAI meets GPU deployment milestones. These shares represent approximately 10% of AMD. The deal also stipulates that this arrangement will only be exercised if AMD’s share price increases due to this groundbreaking partnership.
Industry watchers call it a breakthrough for AMD. It makes AMD a strategic supplier to a leading AI company, which could entice more Tier 1 AI firms. On the other hand, OpenAI is able to lock up sufficient data center capacity to build next-generation AI systems. “It’s hard to overstate how difficult it has become to get enough computing power,” said OpenAI co-founder and CEO Sam Altman.
“This is a major win for AMD and shows that it has been putting the right strategy in place to take advantage of the AI mega-wave with its advanced GPUs,” said Jack Gold, President and Principal Analyst at J. Gold Associates, LLC. “It’s also an indication that OpenAI recognizes a need to diversify its processor supplies, as it continues to expand its data centers.”
Especially when most advanced Nvidia GPUs are on allocation with buildouts outpacing supplies. “By solidifying AMD chip supplies through this commitment and investment, OpenAI can continue its massive buildout campaign,” Gold added.
AMD versus Nvidia
Another upside for AMD in this landmark deal is that it enables the company to challenge Nvidia’s dominance in the AI compute market. “It’s an endorsement that AMD has become competitive with Nvidia’s processing power, as well as enhancing its software strategy to compete, much like it did in the data center market earlier against Intel,” Gold said.
According to Mizuho Securities, Nvidia has captured more than 70% of the market for AI chips. Here, while AMD’s processors are widely used in gaming, PCs, and traditional data center servers, they haven’t made a lot of inroads in pricier compute chips for advanced AI systems. In fact, AMD and other Nvidia rivals have sought to offer more affordable AI chip alternatives as secondary suppliers.
This arrangement between AMD and the maker of ChatGPT could potentially disrupt Nvidia’s AI dominance. “It’s highly likely that other major AI players will follow suit and deploy AMD-powered data center chips,” said Gold. “Ultimately, this is a huge win for AMD and its ability to become a true competitor to Nvidia in AI and the first choice in AI processors.”
It’s important to note that OpenAI’s tie-up with AMD is definitive, unlike its $100 billion deal with Nvidia announced last month, in which Nvidia would become a “preferred strategic compute and networking partner” for building new AI data centers. The Nvidia deal isn’t yet completed; although the two companies have signed a letter of intent, they have yet to disclose details about the regulatory filing.
SemiAnalysis’s Nano acknowledged that the structure of the deals is very different. He also quoted OpenAI chief Altman, “We plan to increase our Nvidia purchasing over time. The world needs much more compute.”
Exciting times in AI
Forrest Norrod, executive VP at AMD, calls this deal transformative, not just for AMD, but for the dynamics of the industry. AMD, striving to establish itself as a credible alternative to Nvidia’s AI hardware and software, has finally scored a significant win. No wonder Lisa Su calls it a major reflection point for AMD.
AMD generated $5 billion from its Instinct GPUs in 2024, compared to Nvidia’s $102.2 billion from data center compute products. That could change with AMD’s compute commitment with OpenAI; it could eventually narrow Nvidia’s AI lead and bring hyperscalers like Meta and Microsoft to AMD’s door.
According to Gold, in two to three years, AI will shift its focus to inference as the primary workload, rather than relying on large-scale data center expansions used for model training. “However, in the meantime, there is a massive market for GPUs, and AMD is now able to capture a significant piece of this revenue.” He also points out that it’s a major win for TSMC, as it fabricates chips for both AMD and Nvidia. “TSMC profits when either AMD or Nvidia powers new data centers.”
Exciting times in the world’s most ambitious AI buildout and advancement of the entire AI ecosystem.
From EEtimes