Photo: Intelligencer; Photo: Getty Images

Nvidia has only attained household-name status recently, but it’s the opposite of an overnight success. For decades, the company was mostly known for designing graphics processor units, or GPUs, for gaming, a market it eventually dominated. More recently, Nvidia’s longtime investment in applications beyond gaming started to bear fruit. First, crypto miners started hoarding Nvidia hardware, giving the company a surprise boost. Then, the company’s early alignment with GPU-hungry AI firms really paid off, turning it into arguably the biggest beneficiary of the recent boom. The global AI-driven data center buildout has pushed Nvidia’s quarterly revenue from around $13 billion when ChatGPT was released to more than $165 billion; currently, by market cap, it’s the most valuable company in the world.

This is all wonderful news for Nvidia founder and CEO Jensen Huang, as well as the company’s shareholders. Things are a bit more complicated for Nvidia’s major customers, whose all-in bets on AI depend on, among many other things, getting sufficient allocations of hardware from a single company with massive pricing power. Google, Amazon, and even Meta have tried to deal with Nvidia’s dominance by investing heavily in their own chips to deploy alongside Nvidia’s while preparing their infrastructure to handle more alternative hardware in the future. The Chinese government sees things similarly, responding to American export limits intended to limit AI progress with a broader ban on Nvidia hardware intended to spur domestic competition. Across the board, these efforts are starting to pay off.

OpenAI is in a particularly weird position. The AI giant is a huge Nvidia customer, spending much of the money it has raised on Nvidia products, or with other firms that have access to Nvidia hardware. As of last month, Nvidia is also a major investor in OpenAI, which will spent at least some of this capital with … Nvidia. The two companies are close and co-dependent but in ways that are far riskier for OpenAI. OpenAI has made some noise about building its own chips and is currently partnering with Broadcom on some custom hardware. But then, this week, it did this:

OpenAI and Advanced Micro Devices announced on Monday a new partnership that would see OpenAI deploy six gigawatts of AMD’s chips over multiple years, starting in the second half of next year. That deal is larger than the $300 billion, 4.5 gigawatt cloud deal OpenAI struck with Oracle earlier this year.

As part of the deal, OpenAI will receive the option to purchase up to 10% of AMD stock if it hits certain milestones.

AMD, Nvidia’s old rival from the graphics-card wars, has been trying for years to get in on the AI boom, and its recent chips have started gaining traction in limited applications. Still, according to insiders, it’s a comparatively small and specialized firm with a lot of catching up to do.

The simplistic framing is

Nvidia gets OAI equity so that OAI can buy more chips

AMD needs to give away their equity to OAI so that they can buy their GPUs

Regardless, congrats AMD today’s your day

— Fabricated Knowledge (@_fabknowledge_) October 6, 2025

Plenty of people have started noting, with some nervousness, the circularity of recent AI deals, but they’re also becoming more speculative. In this case, OpenAI gets access to its stake in AMD if it follows through on spending, and if AMD’s stock price reaches certain milestones (it was up nearly 25 percent on Monday, and Nvidia’s was slightly down). OpenAI’s commitment also gets way out ahead of what AMD can competitively provide now, which means the whole thing is contingent on the company hitting its goals for developing much better hardware as well — something that Nvidia can attest often takes longer than expected. OpenAI isn’t just looking around for an Nvidia competitor, in other words. In AMD, it’s hoping it can help manifest one.

Sign Up for John Herrman column alerts

Get an email alert as soon as a new article publishes.

Vox Media, LLC Terms and Privacy Notice