In a landmark move that underscores the rapidly evolving landscape of artificial intelligence infrastructure, Meta Platforms Inc. has secured a multi-year agreement with Amazon Web Services to utilize its custom Graviton5 processors. The deal, valued in the billions of dollars, will see Meta deploy tens of millions of Graviton5 cores to power its expanding suite of AI services, according to an AWS executive who spoke with Reuters.
The timing of this agreement is strategic. Meta has projected its 2026 capital expenditures to reach between $115 billion and $135 billion, a significant increase from $72.22 billion in 2025. This surge in spending is directed primarily toward Meta Superintelligence Labs and its core business operations as the company doubles down on AI development.
This partnership reflects a broader shift in the AI industry's economic focus—from the training of large models to the operation of AI services at scale. Meta's deployment of Graviton cores will begin with tens of millions of units and will support what the company calls 'agentic AI': systems capable of reasoning, planning, and executing complex tasks with minimal human intervention.
Amazon vice president and distinguished engineer Nafea Bshara characterized the agreement as extending beyond mere chip procurement. For Meta, diversifying its compute options has become a 'strategic imperative,' noted Santosh Janardhan, who oversees infrastructure at the company, as it pushes AI to unprecedented scale.
Importantly, Meta is not abandoning the graphics processing units that initially fueled generative AI. Nvidia GPUs remain central to Meta's large model training, and the company continues to maintain chip agreements with both Nvidia and Advanced Micro Devices. Meta is also collaborating with Arm Holdings on silicon for its data centers, further broadening its technological base.
On the cost front, Meta is implementing significant workforce reductions. The company is cutting approximately 10% of its staff—around 8,000 jobs—and closing 6,000 open positions, according to a memo from Chief People Officer Janelle Gale. The layoffs are effective May 20 and are intended to make Meta 'more efficiently' run, Gale wrote.
Market reaction was positive. Meta shares rose $16.87 to close at $676.02, while Amazon gained $8.87 to $263.95. The stock movements reflect investor optimism about Meta's AI strategy and cost-cutting measures, even as the company faces mounting regulatory scrutiny.
Analysts see Meta at a crossroads. Minda Smiley, senior analyst at eMarketer, described the company as facing a 'crossroads moment' despite solid business performance, with robust ad revenue, sharp AI investment, layoffs, and regulatory challenges all converging simultaneously.
Investors will get their next detailed look at Meta's financials after the market close on April 29, when the company reports first-quarter results. According to S&P Global and Visible Alpha consensus, Meta's Q1 revenue is projected at $55.5 billion, with operating profit of $19.4 billion. Reality Labs, Meta's VR and AR division, is expected to post approximately $5 billion in losses. Capital spending forecasts for 2026 hover near $125 billion.
For Amazon, landing this order validates that its homegrown chips can handle heavy-duty AI workloads, not just cost-effective alternatives. For Meta, the deal adds another supplier to its mix as CEO Mark Zuckerberg pushes for more AI muscle while striving to prevent infrastructure expenses from outpacing growth.
