Friday, June 26, 2026
EN·DarkSubscribe
AI Infrastructure · News & Analysis
HomeCompute & CloudReport
Compute & Cloud · Report

Meta committed over $100 billion to a long-term partnership with AMD for AI infrastructure and computing capability expansion.

Meta's massive investment in AMD signals strategic diversification away from Nvidia dependence and establishes AMD as a viable alternative for hyperscaler capex, reshaping GPU procurement competition.
Trade pressSlicast · February 24, 2026 · Global · Source: ibtimes.com.au
importance 85

Advanced Micro Devices Inc. announced a landmark multi-year partnership with Meta Platforms Inc. on Tuesday, February 24, 2026, to supply up to 6 gigawatts of AMD Instinct GPUs for Meta's next-generation AI infrastructure. The deal, valued at more than $100 billion over five years according to market estimates, represents one of the largest single commitments in the AI accelerator market and positions AMD as a key tier-1 supplier alongside dominant player Nvidia. The announcement sent AMD stock surging as much as 14% in pre-market trading, with shares jumping from around $180 to over $205, adding billions to the company's market cap.

Under the definitive agreement, Meta will deploy AMD Instinct GPUs across multiple generations, beginning with a custom variant based on the MI450 architecture optimized for Meta's workloads. Shipments for the first gigawatt are expected to commence in the second half of 2026, powered by the AMD Helios rack-scale architecture—originally announced at the 2025 Open Compute Project Global Summit—and paired with sixth-generation AMD EPYC "Venice" CPUs running ROCm software. As a strategic incentive, AMD will issue Meta a performance-based warrant for up to 160 million shares of common stock, vesting in tranches tied to stock price milestones and technical and commercial achievements. If fully vested, this could give Meta ownership of roughly 10% of AMD.

Analysts hailed the partnership as transformative. "This is a game-changer," said one semiconductor expert, noting it provides near-term revenue visibility and credibility in hyperscaler contracts. The agreement follows Meta's recent expansion with Nvidia, where the company committed to deploying millions of Blackwell and upcoming Rubin GPUs, underscoring Meta's dual-sourcing strategy and massive AI buildout needs. For AMD, the partnership bolsters its challenge to Nvidia's dominance in AI training and inference, with AMD CEO Lisa Su emphasizing ecosystem collaboration and the ROCm software stack improvements to close the gap with Nvidia's CUDA platform.

AMD stock has risen more than 80% over the past year amid strong demand for Instinct accelerators, though it remains below some fair-value estimates amid broader sector volatility. The deal arrives amid intense scrutiny of AI capex, yet CEO Mark Zuckerberg has defended Meta's aggressive investments as essential for long-term competitiveness. Investors will watch closely for deployment milestones and first shipments in late 2026, as the partnership could add tens of billions in cumulative revenue over the multi-year term, providing a strong tailwind for AMD's data center segment and highlighting the high stakes for chipmakers in the AI arms race.

Read the original
Meta committed over $100 billion to a… · Slicast