Advanced Micro Devices has announced one of the most consequential deals in the AI chip industry’s history. On July 22, 2026, AMD and Anthropic confirmed a multi-part agreement that will see AMD sell Anthropic tens of billions of dollars worth of AI servers, invest up to $5 billion in the Claude maker, and collaborate on optimising AMD’s software stack for Anthropic’s model infrastructure. The deal gives Nvidia’s closest rival its most significant foothold yet in a market the Jensen Huang-led company has dominated since the AI boom began.
Anthropic will buy up to two gigawatts of AMD’s latest-generation Instinct MI450 chips, starting in the first half of 2027. AMD executives have said 1 gigawatt of computing power, enough for roughly 750,000 US homes, can cost double-digit billions of dollars. Two gigawatts at that cost structure implies a hardware commitment in the range of $30 billion or more making this one of the largest chip procurement deals ever announced.
“AMD to sell Anthropic tens of billions in AI servers, invest up to $5 billion in startup in a deal that would strengthen AMD’s foothold in a market dominated by rival Nvidia. Anthropic will buy up to 2 gigawatts of AMD’s Instinct MI450 chips starting in H1 2027.”~Reuters
The Circular Deal Pattern And How This One Differs:
The announcement marks the latest so-called circular deal in the AI industry, where chipmakers invest in AI firms that are among their biggest customers. Nvidia has been in talks to invest $30 billion in ChatGPT creator OpenAI. In October 2025, AMD announced a multi-year deal with OpenAI that would also bring in tens of billions of dollars in annual revenue while giving the ChatGPT creator the option to buy up to roughly 10% of the chipmaker.
The fact that AMD’s equity investment is milestone-linked distinguishes the Anthropic agreement. After meeting defined deployment milestones, AMD will make a strategic investment of up to $5 billion in Anthropic. This is AMD’s first direct investment in Anthropic. AMD’s investment in Anthropic will be dependent on reaching certain deployment objectives. This approach prevents AMD from overcommitting capital before the chips are deployed at scale, providing a hedge against the demand uncertainty that has occasionally alarmed AI infrastructure investors.
The final component of the arrangement, an engineering partnership agreement to optimize AMD’s ROCm software platform for Anthropic’s Claude models, is perhaps as essential as the hardware investment. Nvidia’s dominance is based not only on chip performance, but also on CUDA, a software environment that makes Nvidia chips significantly easier to utilize for AI research than any other. AMD’s ROCm has traditionally struggled to match CUDA’s developer experience. Anthropic’s engineering team is co-developing ROCm optimisations on the MI450 architecture, which is the type of real-world software development that significantly reduces that gap.
“Thrilled to announce our landmark partnership with Anthropic — tens of billions in AI server commitments, up to $5B investment, and joint engineering collaboration on ROCm for Claude. This is a defining moment for AMD’s AI journey and for building a more competitive AI ecosystem.”~Lisa Su
Anthropic’s Compute Scramble: SpaceX, Meta, And Now AMD All In Play
The AMD deal is the latest in a rapid sequence of infrastructure moves by Anthropic to secure the compute it needs to remain competitive with OpenAI and Google DeepMind. Anthropic agreed in May to rent the full computing power of SpaceX’s Colossus 1 facility in Memphis, which houses more than 220,000 Nvidia processors and gave it 300 megawatts of new capacity. Meta Platforms is also in talks to lease computing power to Anthropic in a potential deal worth up to $10 billion over two years.
Under the latest deal, Anthropic would use some of the AMD chips at its own data centers and lease additional capacity through cloud providers and new AI cloud companies. AMD was also in talks to provide a financial backstop for Anthropic’s future data-center leases. That financial backstop role where AMD essentially underwrites Anthropic’s ability to commit to future data centre capacity goes well beyond a chip supply relationship. It makes AMD a financial partner in Anthropic’s infrastructure buildout.
“Access to compute is central to keeping Claude at the frontier and meeting demand from our customers,” Tom Brown, co-founder and compute head at Anthropic, said. The startup has moved aggressively in recent months to overcome capacity constraints for its services.
“Access to compute is central to keeping Claude at the frontier and meeting demand from our customers. The AMD partnership gives us 2GW of MI450 capacity starting H1 2027, engineering collaboration on ROCm, and a long-term supply partner that complements our existing infrastructure.”~Tom Brown, Anthropic
What The Deal Means For AMD’s Nvidia Challenge:
“It’s strategically significant, as each win deepens AMD’s credibility as the number-two to Nvidia and supports the slow, steady share gains underpinning its AI growth story,” said Emarketer analyst Jacob Bourne. Shares of the Santa Clara, California-based company rose 2.4%. The stock has more than doubled in value so far this year.
For AMD, the Anthropic deal is a proof point that its Instinct AI accelerator line can win major frontier model lab commitments not just from OpenAI, which signed its own AMD deal in October 2025, but now from Anthropic as well. Two of the three most prominent frontier AI labs in the world have now committed to AMD hardware at scale, a development that would have seemed improbable two years ago when Nvidia’s GPU supply was the only serious option for training large language models.
Whether AMD can actually deliver 2 gigawatts of MI450 capacity by the first half of 2027 will depend heavily on TSMC’s CoWoS advanced packaging throughput and HBM memory supply both of which remain constrained. But the commercial signal sent by this deal is unambiguous: the era of Nvidia’s near-total monopoly on frontier AI compute is ending, and AMD is the primary beneficiary.
