Amazon Locks Down Meta as Chip Customer While Deepening $100 Billion Anthropic Pact

Amazon has secured a major new customer for its homegrown semiconductor business, with Meta committing to deploy hundreds of thousands of Graviton processors across its next-generation artificial intelligence infrastructure. The multiyear agreement, spanning at least three years, will see the social media giant tap into tens of millions of compute cores running on Amazon’s proprietary chip architecture.

The deal underscores a strategic shift in how hyperscalers are positioning themselves in the AI arms race. While graphics processing units remain the workhorses for training large language models, the rise of agentic AI—autonomous systems that execute complex tasks in real time—is driving surging demand for CPU-based compute. Amazon’s latest Graviton5 chip addresses this directly, delivering 25% more performance than its predecessor with significantly faster inter-core communication. Meta joins a growing roster that already includes Apple, Snowflake, and Anthropic.

The timing of the announcement proved fortuitous for Amazon’s stock. Shares hit a fresh 52-week high of €223.70 on Friday, climbing 2.64% in a single session and extending a monthly rally that has now topped 22%. The move has prompted a flurry of analyst upgrades, with BMO lifting its price target to $315, UBS raising its target to $304 while forecasting 38% AWS growth by 2026, and Oppenheimer setting a $275 target with an outperform rating.

The Meta deal is far from Amazon’s only chip-related coup. The company has simultaneously deepened its already substantial ties with Anthropic, the developer behind the Claude family of AI models. In a blockbuster expansion of their partnership, Anthropic has committed to spending $100 billion on Amazon Web Services cloud infrastructure over the next decade. In exchange, Amazon is injecting an additional $5 billion into the startup, bringing its total investment to $13 billion, with the structure allowing for future capital infusions of up to $25 billion contingent on Anthropic hitting commercial milestones.

Should investors sell immediately? Or is it worth buying Amazon?

Crucially, Anthropic is doubling down on Amazon’s proprietary silicon. The startup will exclusively use Amazon’s Trainium and Inferentia chips for all future model training and inference, abandoning third-party GPU alternatives. A centerpiece of this strategy is “Project Rainier,” a massive compute cluster housing nearly half a million Trainium2 chips—one of the largest AI supercomputers ever built. Amazon is betting that its vertically integrated hardware stack can undercut traditional GPU-based solutions on both cost and efficiency, a value proposition it aims to pass on to enterprise customers accessing Anthropic’s models through the Amazon Bedrock platform.

All eyes now turn to April 29, when Amazon reports first-quarter earnings. Analysts expect revenue of roughly $188 billion, with AWS contributing nearly $37 billion. The market is bracing for capital expenditure guidance that has been revised sharply upward, with 2026 spending projections now approaching $200 billion. The critical question for investors is whether the massive outlays on proprietary chip infrastructure and strategic partnerships are beginning to generate the promised returns—or whether they will weigh on near-term margins. Any sign of softening cloud growth could quickly cool the stock’s recent rally.

Ad

Amazon Stock: New Analysis - 25 April

Fresh Amazon information released. What's the impact for investors? Our latest independent report examines recent figures and market trends.

Read our updated Amazon analysis...