- Meta commits to six gigawatts of AMD Instinct GPU hardware
- Initial implementation of one gigawatt is planned for the second half of 2026
- Custom MI450 silicon developed specifically for Meta AI workloads
AMD and Meta have agreed to deploy up to 6GW of AMD Instinct GPUs over multiple generations, marking one of the largest unveiled infrastructure commitments in the AI sector.
The scale of the plan suggests a major expansion of Meta’s internal computing capacity as demand for training and running advanced AI models continues to grow.
The first phase of shipments is expected to begin in the second half of 2026 and will support an initial 1GW deployment.
First implementation timeline is taking shape
This rollout will rely on a custom GPU derived from AMD’s MI450 architecture and integrated into the company’s Helios rack-scale platform.
The systems will also use sixth-generation EPYC processors, codenamed Venice, along with ROCm software.
The agreement builds on an existing relationship where Meta has already installed a large number of EPYC CPUs and earlier Instinct GPUs across its global data center.
As AI workloads increase in complexity, the balance between CPUs and GPUs has become more critical, especially for orchestration, scheduling and efficiency across massive clusters.
In addition to the hardware supply agreement, the companies are developing a custom GPU optimized for Meta’s AI workloads.
The custom MI450-based design is expected to support large-scale AI training and inference tasks across Meta’s services.
Meta says AI infrastructure is central to delivering services to billions of users, including recommender systems, generative tools and future applications that rely on intensive computation.
“We are excited to form a long-term partnership with AMD to implement efficient inferential computing and deliver personalized superintelligence,” said Mark Zuckerberg, founder and CEO of Meta.
“This is an important step for Meta as we diversify our computing. I expect AMD to be an important partner for many years to come.”
For AMD, the deal deepens its presence in hyperscale data centers at a time when competition in AI accelerators remains intense.
“We are proud to expand our strategic partnership with Meta as it pushes the boundaries of artificial intelligence on an unprecedented scale,” said Dr. Lisa Su, Chairman and CEO, AMD.
“This multi-year, multi-generational collaboration across Instinct GPUs, EPYC CPUs and rack-scale AI systems aligns our roadmaps to deliver high-performance, energy-efficient infrastructure optimized for Meta’s workloads, accelerates one of the industry’s largest AI deployments, and places AMD at the center of global AI deployment.”
The partnership also includes a performance-based warrant that allows Meta to acquire up to 160 million AMD shares, with vesting tied to shipment and stock milestones.
The first tranche is earned after the initial gigawatts of GPU shipments, while additional tranches depend on scaling purchases towards the full 6GW commitment.
Such structures link commercial success directly to equity participation and adjust incentives as the infrastructure expands.
“We expect this partnership to drive significant revenue growth over several years and be accretive to our non-GAAP earnings per share,” said Jean Hu, EVP, CFO and Treasurer, AMD.
“The performance-based structure also closely aligns AMD and Meta around execution and long-term value creation.”
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews and opinions in your feeds. Be sure to click the Follow button!
And of course you can too follow TechRadar on TikTok for news, reviews, video unboxings, and get regular updates from us on WhatsApp also.
