Why is Meta Partnering with AMD to Power AI Infrastructure?

Meta and Advanced Micro Devices (AMD) have announced their partnership to power AI infrastructure for the next generation of artificial intelligence.

The agreements aligns with Meta's AI demands, and AMD's goals to be at the centre of AI construction and deployment across the globe.

The aim is the buildout of up to 6GW of AMD's Instinct GPUs for Meta's growing workloads, which claims it requires scalable compute power for its worldwide operations.

Mark Zuckerberg, Founder and CEO of Meta, says: "We're excited to form a long-term partnership with AMD to deploy efficient inference compute and deliver personal superintelligence.

"This is an important step for Meta as we diversify our compute. I expect AMD to be an important partner for many years to come," he adds.

AI bubble fears

The partnership between Meta and AMD is worth US$60bn of investment for Meta's CEO, and US tech companies are expected to spend US$660bn on AI assets, according to Alvin Nguyen, an analyst at Forrester.

An estimated US$3tn is projected for data centre spend by AI giants like OpenAI, Google and AI chipmaker NVIDIA, which are some of the world's most valuable companies.

The partnership comes at a time where progress of artificial general intelligence (AGI) could stall as investors expect a return on their trillions, while there is the potential risk of AI companies fall short and causing US stock markets to plummet. 

However, as NVIDIA became the first US$5tn company nearing the end of last year, there may currently be few signs of a burst. A US$60bn deal by Meta carries its ROI risks, as billion-dollar investments will – but as of right now, company value and revenue continues to grow.

AI at hyperscale

Meta says its collaboration with AMD across both software and hardware will enable them to innovate quickly and at scale.

The first GPU shipments will begin in the second half of 2026, Meta writes. They will be built on the Helios rack-scale architecture, a type of open source hardware which aims to "make data centre infrastructure more efficient, scalable, and sustainable."

Dr. Lisa Su, Chair and CEO at AMD, says: "We are proud to expand our strategic partnership with Meta as they push the boundaries of AI at unprecedented scale.

"This multi-year, multi-generation collaboration across Instinct GPUs, EPYC CPUs and rack-scale AI systems aligns our roadmaps to deliver high-performance, energy-efficient infrastructure optimised for Meta's workloads, accelerating one of the industry's largest AI deployments and placing AMD at the centre of the global AI buildout."

Meta's data centre plans

Meta will invest over US$600 billion by 2028 to build "gigawatt-plus" scale campuses across US states, including Louisiana, Texas, Wisconsin, Wyoming and more. There are plans for hyperscale data centres that are capable of handling AI-heavy workloads:

The first multi-gigawatt AI data centre, called Prometheus Hyperscale, is expected to come online in 2026 in Ohio. Another data centre, called Hyperion, could scale up to five gigawatts over several years.

In a post on his social media platform, Threads, Mark Zuckerberg said that one data centre site would cover an area nearly the size of Manhattan (22.8sq miles).

Meta Compute initiative

Meta says its agreement with AMD is part of its Meta Compute initiative, which aims to scale infrastructure for "the era of personal superintelligence, future-proofing our leadership in AI".

Concluding its aims for the project, Meta says: "We believe this portfolio approach will enable us to advance and innovate at an unmatched pace, rolling out powerful, efficient new hardware co-designed with our software stack to handle massive growth.

"We look forward to working with AMD to power our AI innovations and secure our ability to deliver world-class AI experiences to billions of people globally."