Key Highlights
- Advanced Micro Devices has finalized a massive agreement with Anthropic spanning tens of billions in AI server infrastructure powered by AMD’s forthcoming Instinct MI450 processors
- The AI startup will acquire up to 2 gigawatts worth of processing chips, with shipments commencing during H1 2027
- In its inaugural equity stake in the AI company, AMD commits up to $5 billion to Anthropic
- AMD shares experienced volatility, initially rising in pre-market before declining approximately 3% amid broader semiconductor sector weakness
- Discussions are underway for AMD to guarantee financing for Anthropic’s upcoming data center lease obligations
In what represents one of the semiconductor industry’s most substantial agreements in recent times, Advanced Micro Devices and Anthropic have forged a partnership encompassing tens of billions of dollars in AI server infrastructure centered on AMD’s next-generation Instinct MI450 processors.
The Claude AI developer has committed to acquiring up to 2 gigawatts of computing power through these advanced chips, with the rollout scheduled to commence in early 2027. AMD shares showed modest gains of approximately 0.88% during regular trading, though the stock experienced pre-market volatility—briefly falling 3% after an initial surge dissipated amid a wider technology sector downturn.
Advanced Micro Devices, Inc., AMD
Marking AMD’s maiden equity position in Anthropic, the semiconductor manufacturer has pledged up to $5 billion in funding tied to specific deployment benchmarks. This substantial financial backing demonstrates AMD’s determination to establish itself as a critical player in artificial intelligence infrastructure.
AMD’s Chief Executive Lisa Su spoke openly about the company’s strategic objectives. “We have very much wanted to be a major part of their infrastructure,” Su explained, noting that technical teams from both organizations have been collaborating extensively behind the scenes for an extended period.
From Anthropic’s perspective, this partnership solves a critical operational challenge. The company has faced capacity constraints due to explosive growth in demand for its AI products, with certain customers experiencing service interruptions and usage restrictions implemented to balance system loads.
The AI company currently operates workloads on diverse hardware platforms—including Google’s Tensor Processing Units, Amazon’s Trainium processors, and Nvidia’s graphics processing units. This new partnership will enable Anthropic to install AMD silicon in proprietary data facilities while also accessing additional computing resources through hyperscale cloud partners and specialized infrastructure providers.
Su emphasized the extensive planning required for infrastructure projects of this magnitude. “You can’t just wake up one morning and say, ‘Oh, I want a gigawatt of compute tomorrow,'” she noted. “You actually have to plan 12, 18, 24 months in advance.”
AMD Intensifies Competition With Nvidia
This landmark agreement represents AMD’s continued campaign to capture market share from Nvidia’s commanding position in AI accelerator hardware. The company has been securing customers by capitalizing on AI developers’ desire to diversify their supplier base beyond a single vendor.
AMD’s recent successes include significant contracts with OpenAI and Meta Platforms. The Anthropic partnership now adds another elite AI organization to AMD’s growing client portfolio.
To secure these partnerships, AMD has deployed creative incentive structures, including providing equity warrants to customers. This approach appears to be yielding results as additional AI enterprises explore alternatives to Nvidia’s platform.
AMD Considers Financial Support for Anthropic’s Infrastructure Expansion
Extending beyond chip supply, AMD is negotiating terms to provide financial guarantees for Anthropic’s future data center lease commitments, according to sources familiar with the negotiations.
This arrangement aligns with an emerging industry pattern where established technology corporations with strong credit ratings enable AI startups to obtain infrastructure financing they would otherwise struggle to secure independently. Google has previously supported certain Anthropic data center agreements connected to its TPU hardware supply.
The comprehensive partnership also includes a technical collaboration component where Anthropic’s Claude language models will contribute to optimizing AMD’s processor capabilities.
Earlier this year, Anthropic expanded its cloud infrastructure partnerships with both Google and Amazon, while also striking an arrangement with Elon Musk’s SpaceX, which has been monetizing surplus data center capacity from its infrastructure investments.



