AMD is making a big bet on Anthropic, the AI company best known for its coding assistant Claude Code. The chipmaker intends to put as much as $5 billion into Anthropic. Under the agreement, Anthropic will deploy 2 gigawatts of computing power using AMD Instinct MI450 GPUs in AMD Helios rack-scale solutions.
One gigawatt of that computing power is scheduled to become operational during the first six months of next year. AMD CEO Lisa Su said the company is "thrilled to deepen our partnership with Anthropic and deploy AMD Helios at gigawatt scale."
Anthropic's products have grown faster than its infrastructure can handle. In April, the company acknowledged that the surge in usage of its Claude models and products had placed its infrastructure under "inevitable strain," degrading "reliability and performance" for users, especially during peak hours.
So Anthropic is signing deals everywhere it can. The company already has multi-billion-dollar partnerships with Amazon and multi-gigawatt compute agreements with Google and Broadcom. Anthropic also struck a deal with SpaceX to use all the computational power at the Memphis, Tennessee-based Colossus 1 data center.
Get the market news that matters in a five-minute read with Market Briefs, our free daily newsletter
Under that arrangement, Anthropic is obligated to pay SpaceX $1.25 billion each month until May 2029, according to SpaceX's prospectus. And it is in early-stage discussions about renting compute capacity from Meta.
The numbers behind Anthropic show why it can afford all this. The company generated roughly $10 billion in revenue over the entirety of last year. In May, its run-rate revenue topped $47 billion.
The company's valuation in that same funding round hit $965 billion.
AMD is the main competitor to Nvidia in the GPU market, and Nvidia has dominated the AI chip boom so far. This deal gives AMD a major win in landing a high-profile customer at massive scale. It also comes with a side note: OpenAI, Anthropic's primary rival, already possesses a warrant for 160 million AMD common shares - representing about one-tenth of the firm - and the right to exercise those shares depends on deployment milestones and AMD's share price.
So AMD is selling chips to both sides of the AI arms race. Anthropic gets the hardware it needs to keep its models running. OpenAI gets a potential stake in AMD's future.
Anthropic's monthly SpaceX bill alone is over a billion dollars. The company is readying a massive IPO that could arrive as soon as this year and confidentially filed its prospectus with the SEC in June. That would give public investors a chance to own a piece of the story.
For investors, this deal is a reminder that the AI boom is not just about one chipmaker or one chatbot. The demand for computing power is so intense that companies like Anthropic are signing up for years of capacity in advance. That creates winners across the whole supply chain - chip designers, data center builders, and the companies that run them.
AMD is betting it can take a bigger slice of that spending away from Nvidia. If Anthropic's growth continues and its IPO goes well, the $5 billion investment could look cheap. If the AI market slows or Anthropic stumbles, AMD still gets the revenue from selling those GPUs.
Either way, the race to build the biggest AI infrastructure is just getting started. And your portfolio has a front-row seat to see who wins.
Join Market Briefs, our free daily newsletter, for a quick daily rundown of the markets
