SANTA CLARA, Calif. — Advanced Micro Devices and Meta Platforms announced a multi-year agreement on Tuesday under which Meta will deploy up to 6 gigawatts of AMD’s Instinct graphics processing units, one of the largest chip commitments yet in the race to build artificial intelligence infrastructure.
The deal deepens an existing relationship between the two companies and gives AMD a major anchor customer as it competes with Nvidia, the dominant supplier of the processors used to train and run AI models.
Deployment Timeline
According to the companies, the first gigawatt of capacity is expected to begin shipping in the second half of 2026. That initial deployment will use a custom GPU based on AMD’s MI450 design, paired with the company’s next-generation EPYC processors, code-named Venice, and its ROCm software platform. The systems will be built on AMD’s Helios rack-scale architecture, which packages chips, networking and cooling into integrated data center units.
Measuring the agreement in gigawatts, rather than in numbers of chips, reflects how AI companies increasingly plan their infrastructure around the electric power their data centers can draw. A gigawatt is roughly the output of a large power plant.
Warrant Tied to Milestones
As part of the agreement, AMD issued Meta a warrant to buy up to 160 million AMD shares at an exercise price of $0.01 per share, according to a regulatory filing. The warrant vests in tranches linked to Meta’s purchases and AMD’s shipments of GPUs, scaling from the first gigawatt up to the full 6 gigawatts.
Vesting is also tied to AMD’s share price reaching specified thresholds, the highest of which is $600, and to Meta meeting certain technical and commercial conditions. The warrant expires on Feb. 23, 2031, the filing showed.
The structure mirrors arrangements in which chip suppliers offer equity incentives to secure long-term demand from the largest AI buyers, aligning the customer’s interests with the supplier’s stock performance.
Meta’s Spending Spree
Meta has committed vast sums to computing capacity as it develops its own AI models and builds features across Facebook, Instagram and WhatsApp. The company has been diversifying its chip suppliers, buying heavily from Nvidia while also developing custom silicon and expanding its work with AMD.
For AMD, the agreement provides multi-year visibility into demand for its data center accelerators at a time when investors are closely watching whether the company can capture a larger share of the AI chip market. Its Instinct line has gained traction with large cloud and internet companies seeking alternatives to Nvidia’s hardware.
Industry Context
The announcement adds to a series of multi-gigawatt infrastructure deals struck by technology companies over the past year, as they race to secure the chips, power and data center space needed for increasingly large AI systems. Analysts have raised questions about whether such spending can be sustained, while the companies argue that demand for AI services justifies the investment.
Neither company disclosed the total dollar value of the agreement.
Sources & Credits
- AMD Investor Relations: AMD and Meta Announce Expanded Strategic Partnership to Deploy 6 Gigawatts of AMD GPUs
- AMD Form 8-K (SEC filing): Form 8-K, February 24, 2026
Image: Coolcaesar, CC BY-SA 4.0, via Wikimedia Commons.




Join the conversation