- AMD will invest up to $5 billion in Anthropic equity, contingent on deployment milestones, with Anthropic deploying up to 2GW of Instinct MI455X GPUs via AMD's Helios rack-scale systems [1] - The first gigawatt of compute begins deployment in H1 2027, pairing MI455X GPUs with AMD EPYC "Venice" CPUs and Pensando networking [1] - Unlike AMD's prior deals with Meta and OpenAI — which included equity warrants — the Anthropic agreement contains no warrant provisions, signaling a shift in chipmaker-AI company dynamics [3] - AMD and Anthropic will pursue a multi-year engineering collaboration, using Claude to optimize workloads on AMD's ROCm software stack and Instinct GPUs [1] - The deal brings AMD's total announced AI compute commitments to roughly 14GW across Anthropic, Meta, and OpenAI
[4] AMD announced on July 22 that it will invest up to $5 billion in Anthropic as part of a sweeping infrastructure partnership that will see the AI company deploy up to 2 gigawatts of AMD's next-generation Instinct MI450 Series GPUs. The deal, unveiled at AMD's "Advancing AI 2026" event, pairs Instinct MI455X accelerators with EPYC "Venice" CPUs and Pensando networking inside AMD's new Helios rack-scale architecture [1].
Deployment of the first gigawatt begins in the first half of 2027, with Anthropic acquiring chips through a mix of direct ownership and leased capacity from cloud and neocloud providers. AMD is also exploring financial backing for some of Anthropic's future data-center lease obligations [3].
The agreement notably lacks the equity warrants AMD extended to Meta and OpenAI in earlier deals — Meta received a performance-based warrant for up to 160 million AMD shares — suggesting AMD's bargaining position has strengthened as AI compute demand continues to outstrip supply [4].
The Deal #
Under the terms announced Tuesday, AMD's $5 billion equity investment in Anthropic is structured as a commitment tied to deployment milestones rather than a lump-sum payment. Anthropic will deploy the MI450 Series chips in its own facilities and through third-party cloud infrastructure, with AMD collaborating to identify suitable data center locations [1].
Anthropic already runs workloads on AMD's prior-generation Instinct MI355X GPUs alongside chips from Google, Amazon, and Nvidia. The new partnership substantially deepens that relationship, making AMD a primary hardware supplier for Claude's training and inference infrastructure [1].
The companies also announced a multi-year engineering collaboration: Anthropic will use Claude to optimize workloads for AMD Instinct GPUs and accelerate development of AMD's ROCm software stack, while AMD will broadly adopt Claude across its own engineering and product development teams [1].
Why It Matters #
The Anthropic deal brings AMD's total announced AI compute commitments to approximately 14 gigawatts across three frontier AI companies. Meta and OpenAI each signed 6GW agreements for AMD MI450-based infrastructure, with initial 1GW deployments from both expected in the second half of 2026 [4].
For AMD, the traction represents a material challenge to Nvidia's dominance in AI training hardware. Nvidia's CUDA software ecosystem has long been considered a structural moat, but AMD's willingness to invest billions in customer equity and collaborate on software optimization signals a strategy of buying its way into parity. The absence of equity warrants in the Anthropic deal is particularly notable. Barclays analyst Tom O'Malley characterized it as evidence of 'which direction the puck is moving in a supply-constrained world,' arguing that chipmakers no longer need to offer equity sweeteners to secure long-term agreements [3].
Market Reaction #
AMD shares rose approximately 2% in midday trading on July 22 after initial pre-market volatility. Wells Fargo analyst Aaron Rakers maintained an Overweight rating with a $615 price target, noting the deal bolsters AMD's projected data center GPU revenue, which the firm estimates could reach $40.6 billion by 2027 [5].
The stock had initially dipped as much as 3% in pre-market trading amid a broader tech selloff before recovering. AMD shares traded near $530 at the time of the announcement [3].
Competitive Landscape #
AMD CEO Lisa Su framed the Anthropic partnership as the culmination of months of behind-the-scenes engineering work. 'We have very much wanted to be a major part of their infrastructure,' Su said, adding that data center buildouts require planning '12, 18, 24 months in advance' [3].
Anthropic's decision to diversify its chip suppliers beyond Google's TPUs, Amazon's Trainium, and Nvidia's GPUs reflects a broader industry trend. As frontier AI labs race to secure compute for increasingly large training runs, multi-vendor strategies reduce supply-chain risk and improve negotiating leverage.
The Helios rack-scale system — co-developed with Meta through the Open Compute Project — is central to AMD's pitch. By offering a full-stack solution that includes GPUs, CPUs, networking, and software, AMD is positioning itself as more than a chip vendor [4].
Companies mentioned #
Further sources #
[1] AMD and Anthropic Announce Strategic Partnership to Deploy Up to 2 Gigawatts of… ↗
[[2] AMD commits up to $5 billion to Anthropic — The Verge, July 22, 2026 ↗](https://www.theverge.com/ai-artificial-intelligence/969285/amd-anthropic-ai-infrastructure-deal)
[[3] AMD challenges Nvidia with massive Anthropic deal — Yahoo Finance / Quartz, Jul… ↗](https://ca.finance.yahoo.com/news/amd-challenges-nvidia-massive-anthropic-131409608.html)
[4] AMD Stock Near $510 Before Its Biggest AI Event — 12 Gigawatts on the Line — Tr… ↗
[5] AMD Secures Major Deal with Anthropic, Boosting Revenue Projections — GuruFocus… ↗ The stories that matter, in one email. Free — unsubscribe anytime.