In a strategic maneuver aimed at reshaping the semiconductor landscape, AMD has unveiled plans to invest up to $5 billion in Anthropic. The deal, contingent upon the achievement of specific technical and operational milestones, is paired with a massive commitment from Anthropic to purchase tens of billions of dollars worth of AI processors.
A Race for Computational Scale
Anthropic, the creator of the Claude AI model, has committed to securing up to 2 gigawatts (GW) of computing power powered by AMD Instinct MI450 chips. The initial phase, involving 1 GW of capacity, is slated to go online in the first half of 2027. To put this scale into perspective, 1 GW is comparable to the output of a large nuclear reactor, highlighting the immense energy and hardware requirements of frontier AI models.
Breaking the Nvidia Monopoly
This agreement represents a significant victory for AMD as it seeks to erode Nvidia’s dominance in the AI accelerator market. For Anthropic, the move is about supply chain diversification; the company already utilizes hardware from Nvidia, Amazon, and Google. Tom Brown, Anthropic’s co-founder and head of computing infrastructure, noted that access to such power is "critical" for meeting customer demand and maintaining a competitive edge.
Deployment Across Infrastructure
The AMD processors will be deployed across various environments, including Anthropic’s own data centers, third-party cloud providers, and specialized "neocloud" companies. AMD’s growing portfolio of high-scale partnerships, which already includes Meta and OpenAI, positions the company as the primary alternative for large-scale AI development in an era of explosive demand.
