Tech

Meta inks AMD deal to deploy up to 6 GW of GPUs after Nvidia pact

Meta has signed a multiyear deal with AMD to deploy up to 6 gigawatts of AMD GPUs for its AI data centers, plus AI-optimized CPUs. The arrangement comes days after Meta’s expanded AI chip pact with Nvidia, underscoring a broader multi-vendor race to scale compute for large models.

Meta inks AMD deal to deploy up to 6 GW of GPUs after Nvidia pact

Key Takeaways

  • Up to 6 GW of AMD GPUs and AI-optimized CPUs planned for Meta’s AI data centers under a multiyear agreement.
  • Early shipments of AMD MI450 GPUs via Helios rack-scale servers expected later this year.
  • Terms are undisclosed; analyst Ben Bajarin estimates a tens-of-billions-dollar spend over at least four years.
  • Signals a multi-vendor compute strategy at scale, potentially affecting Nvidia pricing and market dynamics.
  • OpenAI previously struck a multiyear AMD deal in October, highlighting AMD’s broader push into AI infrastructure.

People Involved

  • Ben Bajarin Analyst

Entities Involved

  • Meta Platforms, Inc. Social media giant building AI data centers
  • Advanced Micro Devices, Inc. (AMD) AI GPU supplier and processor vendor
  • NVIDIA Corporation Leader in AI GPUs; partner to Meta in earlier pact
  • OpenAI AI research and deployment partner

MarketMoodz Analysis

For investors, the deal broadens Meta’s compute sourcing beyond Nvidia, reducing single-vendor risk and potentially tempering Nvidia’s pricing leverage. The scale—up to 6 GW—points to a substantial long-run capex and a long data-center buildout horizon that could benefit AMD on a revenue and margin basis, if pricing and supply align with Meta’s plans.

Historically, hyperscalers have blended suppliers to optimize for workload type and efficiency. Nvidia has dominated the AI accelerator market, but multi-vendor contracts like this reflect a broader shift toward workload-specific accelerators and rack-scale deployments. Still, several details remain unverified, and the market implications hinge on official disclosures of terms, timelines, and the actual performance of MI450s in Meta’s environments.

What to watch next: seek confirmation of the deal’s terms and pricing from primary sources, monitor AMD’s shipment cadence for MI450 GPUs in Helios servers, and track any changes in Nvidia’s pricing or supply commitments as Meta expands its AI footprint.

Get AI-Powered Market Insights

Stay ahead of market-moving events with our real-time analysis and stock ratings.

Start Your Free Trial