electronics-journal.com
25
'26
Written on Modified on
AMD, Meta Sign 6-GW AI Infrastructure Agreement
Multi-year deployment of AMD Instinct GPUs and EPYC CPUs will power rack-scale AI systems beginning in 2026.
www.amd.com
AI Generated Image
In the AI data center infrastructure and hyperscale cloud computing sectors, AMD and Meta have signed a definitive multi-year, multi-generation agreement to deploy up to 6 gigawatts of AMD Instinct GPUs to support Meta’s next generation of AI platforms. The first deployment, representing one gigawatt of capacity, is scheduled to begin in the second half of 2026.
Rack-Scale Architecture for Gigawatt-Scale AI
The initial deployment will be based on a custom AMD Instinct GPU derived from the MI450 architecture, optimized specifically for Meta’s AI workloads. The systems will operate within the AMD Helios rack-scale architecture, jointly developed by AMD and Meta through the Open Compute Project to enable scalable AI infrastructure at rack level.
The first gigawatt deployment will combine:
In the AI data center infrastructure and hyperscale cloud computing sectors, AMD and Meta have signed a definitive multi-year, multi-generation agreement to deploy up to 6 gigawatts of AMD Instinct GPUs to support Meta’s next generation of AI platforms. The first deployment, representing one gigawatt of capacity, is scheduled to begin in the second half of 2026.
Rack-Scale Architecture for Gigawatt-Scale AI
The initial deployment will be based on a custom AMD Instinct GPU derived from the MI450 architecture, optimized specifically for Meta’s AI workloads. The systems will operate within the AMD Helios rack-scale architecture, jointly developed by AMD and Meta through the Open Compute Project to enable scalable AI infrastructure at rack level.
The first gigawatt deployment will combine:
- Custom AMD Instinct MI450-based GPUs
- 6th Gen AMD EPYC CPUs, codenamed “Venice”
- ROCm software stack
This integrated architecture is designed to support high-density AI training and inference environments at gigawatt scale.
Alignment Across GPU and CPU Roadmaps
The agreement expands the companies’ collaboration beyond GPUs to include AMD EPYC processors. Meta has deployed multiple generations of AMD EPYC CPUs and AMD Instinct MI300 and MI350 GPUs across its global infrastructure.
Under the expanded roadmap alignment, Meta will serve as a lead customer for 6th Gen AMD EPYC processors, including “Venice” and “Verano,” the latter optimized for workload-specific performance-per-dollar-per-watt metrics. CPUs remain a central component of AI infrastructure, providing orchestration, data management and system-level efficiency alongside GPUs.
Structured Multi-Generation Deployment
The partnership includes a performance-based warrant structure under which AMD has issued Meta a warrant for up to 160 million shares of AMD common stock. Vesting is tied to shipment milestones of Instinct GPUs, beginning with the first gigawatt and scaling through the full 6-gigawatt deployment, as well as stock price thresholds and predefined technical and commercial milestones.
This structure aligns both companies around execution targets and long-term deployment objectives.
Strategic AI Infrastructure Expansion
The 6-gigawatt agreement represents a large-scale expansion of AI compute capacity intended to support Meta’s development and deployment of advanced AI models. By aligning silicon, systems and software roadmaps, the collaboration aims to deliver purpose-built AI platforms optimized for hyperscale workloads.
Shipments supporting the first gigawatt of deployment are expected to commence in 2H 2026, marking the beginning of a phased, multi-generation infrastructure rollout.
www.amd.com
Alignment Across GPU and CPU Roadmaps
The agreement expands the companies’ collaboration beyond GPUs to include AMD EPYC processors. Meta has deployed multiple generations of AMD EPYC CPUs and AMD Instinct MI300 and MI350 GPUs across its global infrastructure.
Under the expanded roadmap alignment, Meta will serve as a lead customer for 6th Gen AMD EPYC processors, including “Venice” and “Verano,” the latter optimized for workload-specific performance-per-dollar-per-watt metrics. CPUs remain a central component of AI infrastructure, providing orchestration, data management and system-level efficiency alongside GPUs.
Structured Multi-Generation Deployment
The partnership includes a performance-based warrant structure under which AMD has issued Meta a warrant for up to 160 million shares of AMD common stock. Vesting is tied to shipment milestones of Instinct GPUs, beginning with the first gigawatt and scaling through the full 6-gigawatt deployment, as well as stock price thresholds and predefined technical and commercial milestones.
This structure aligns both companies around execution targets and long-term deployment objectives.
Strategic AI Infrastructure Expansion
The 6-gigawatt agreement represents a large-scale expansion of AI compute capacity intended to support Meta’s development and deployment of advanced AI models. By aligning silicon, systems and software roadmaps, the collaboration aims to deliver purpose-built AI platforms optimized for hyperscale workloads.
Shipments supporting the first gigawatt of deployment are expected to commence in 2H 2026, marking the beginning of a phased, multi-generation infrastructure rollout.
www.amd.com
