Meta Bets on Amazon (AMZN) CPUs as AI Agent Demand Surges

TLDR

• Meta and Amazon struck a multibillion-dollar, multi-year deal for Meta to use AWS Graviton CPU cores for AI.
• The deal covers tens of millions of Graviton cores, mostly located in the U.S., over three to five years.
• Meta’s stock rose 0.6% and Amazon’s rose 1.4% in premarket trading following the announcement.
• The deal adds Amazon to Meta’s existing chip supplier list, which includes Nvidia, Broadcom, and AMD.
• The agreement makes Meta one of AWS’s five largest Graviton customers.

Meta and Amazon have agreed to a multibillion-dollar deal that will see Meta use tens of millions of Amazon Web Services’ Graviton CPU cores to power its AI agent efforts.

The deal runs between three and five years, according to Nafea Bshara, an Amazon VP and co-founder of Annapurna Labs, AWS’s in-house chip unit. Most of the Graviton cores will be deployed in the U.S.

Amazon’s stock gained 1.4% in premarket trading after the news broke, while Meta climbed 0.6%.

Amazon.com, Inc., AMZN

The agreement is specifically for AWS Graviton central processing units — not GPUs. CPUs had taken a back seat as AI development focused heavily on GPU compute, but that’s changing.

The rise of AI agents is bringing CPUs back into focus. They handle specific tasks and feed activity to GPUs, making the two chip types complementary for a wide range of AI workloads.

CPUs also play a key role in the “post-training” phase of large language models, where pretrained models are refined toward specific goals.

Meta chose Amazon’s Graviton5, a 3-nanometer chip, citing its price-to-performance ratio. “Meta has access to so many options from the supply side. But they chose Graviton5,” Bshara said.

Meta’s Diversified Chip Strategy

This deal adds Amazon to a supplier list that already includes Nvidia, Broadcom, AMD, and Arm Holdings. Meta has made clear it doesn’t want to rely on any single chip architecture.

The first deployment will begin with tens of millions of Graviton cores, with room to scale as Meta’s AI needs grow.

Meta’s AI Push and Cost Cuts

Meta’s AI ambitions run deep. The company acquired AI startup Manus for more than $2 billion in December. Manus builds AI agents capable of complex tasks, driving further CPU demand.

To help fund its AI buildout, Meta also announced Thursday it will cut roughly 10% of its workforce — around 8,000 employees — in May.

Meta released its first new AI model in a year, Muse Spark, earlier this month, and has flagged more releases ahead.

The Meta-AWS relationship dates back to around 2016, but had mostly involved cloud services, the Bedrock platform, and GPU cluster rentals. This deal marks a clear expansion into custom silicon.

For AWS, landing Meta as a Graviton customer is a meaningful endorsement. Earlier this week, Amazon also announced a $5 billion investment in Anthropic, which similarly involves tens of millions of Graviton CPU cores.

AWS has been designing its own chips since before 2018, when the first Graviton launched on Arm architecture.

The deal places Meta among AWS’s five largest Graviton customers, Bshara confirmed.

The post Meta Bets on Amazon (AMZN) CPUs as AI Agent Demand Surges appeared first on CoinCentral.