Meta's Dual-Source Chip Strategy Powers AI Ambitions

Meta signs a $60B, 5-year deal with AMD for up to 6 gigawatts of GPU power, securing AI chip supply alongside its recent Nvidia agreement.

In a strategic move to secure the vast computing resources required for its artificial intelligence roadmap, Meta Platforms has entered into a significant new partnership. This follows closely on the heels of another major supply agreement, signaling a deliberate shift away from reliance on a single vendor for critical hardware.

Securing Capacity with AMD

The social media and technology giant announced a multi-year collaboration with Advanced Micro Devices (AMD) this week. The cornerstone of the deal involves Meta deploying up to 6 gigawatts of computing power derived from AMD Instinct GPUs. Industry reports suggest the five-year agreement could be worth as much as $60 billion.

Meta will utilize custom Instinct GPUs built on AMD's MI450 architecture, which are being specifically tailored to the company's workloads. The arrangement also includes AMD's 6th Generation EPYC CPUs (codenamed "Venice") and its forthcoming "Verano" processor.

According to AMD, initial deliveries for a 1-gigawatt deployment phase are scheduled to commence in the second half of 2026. This infrastructure will be based on a rack-scale design known as "Helios," co-developed by both companies through the Open Compute Project.

Innovative Deal Structure Includes Performance Incentives

A notable feature of this partnership is its financial structure. As reported, AMD has granted Meta a performance-based warrant covering up to 160 million AMD shares—approximately 10% of the company—with an exercise price of $0.01 per share.

This warrant will vest in tranches, contingent upon Meta meeting specific milestones. These include shipment volumes, stock price thresholds for AMD, and the achievement of predefined technical and commercial objectives. This mechanism mirrors a similar structure AMD employed in an agreement with OpenAI in October 2025.

A Two-Pronged Approach to Supply

The AMD pact arrives just one week after Meta disclosed a separate, multi-year partnership with Nvidia on February 17. Under that agreement, Meta plans to deploy "millions" of Blackwell and Rubin GPUs, alongside Grace CPUs and networking components, across its hyperscale data centers.

Meta's infrastructure chief, Santosh Janardhan, explained that the scale of the company's expansion necessitates multiple chip suppliers, suggesting that ultimately all major semiconductor makers would have a role to play.

Financial and Legal Headwinds

The aggressive investment pace is reflected in Meta's capital expenditure guidance. In late January, the company projected a 2026 CapEx range of $115 to $135 billion, a substantial increase from the $72.2 billion spent in 2025. Despite reporting robust revenue (Q4 2025: $59.9 billion; Full Year 2025: $201 billion), Meta's stock performance in 2026 has, according to the source material, trailed the broader market.

Legal challenges also persist. On February 18, CEO Mark Zuckerberg provided testimony in a high-profile Los Angeles court case concerning social media addiction, a proceeding viewed as a bellwether for approximately 1,600 similar pending lawsuits.

Currently, the equity appears range-bound. Trading at €544.80, the shares sit slightly below their 50-day moving average and show a decline of 12.86% over a 12-month period.

The critical test will be whether Meta can successfully activate this massive infrastructure offensive on schedule in the latter half of 2026. At that point, the enormous investments must translate into tangible, measurable advances in computational capability and product development.