San Francisco, 24 April 2026 – Meta Platforms has signed a multiyear, multibillion dollar agreement with Amazon Web Services to use Amazon’s in house Graviton chips, marking another major step in the global race to secure computing power for artificial intelligence.
The deal gives Meta access to tens of millions of AWS Graviton central processing unit cores, strengthening the company’s AI infrastructure as it expands work on AI agents, large scale models and related services. The agreement is expected to be worth billions of dollars, according to an AWS executive cited in reports.
For Meta, the move reflects a practical reality in the AI economy. The company cannot rely on one chip supplier or one type of processor alone as it scales its ambitions across social media, advertising, messaging, virtual assistants and enterprise facing AI tools. While graphics processing units remain critical for training large AI models, central processing units are also important for running broader AI workloads, including post training tasks and AI agent operations.
The agreement also gives Amazon a stronger position in the AI chip race. AWS has spent years developing its own silicon, including the Graviton line of CPUs and the Trainium chips used for AI training. By securing Meta as a major customer for Graviton, Amazon can make a stronger case that its custom chips are not only internal cloud infrastructure tools, but also credible alternatives for the world’s largest technology companies.
The reported deal focuses on Graviton5, Amazon’s 3 nanometre chip, which AWS has positioned around price performance advantages. Most of the Graviton cores for Meta are expected to be located in the United States, with the agreement believed to run for several years.
The timing is significant. Global demand for AI infrastructure has surged as technology companies race to develop more powerful models and commercialise AI assistants, chatbots and autonomous software agents. The boom has created intense competition for chips, cloud capacity, data centre power and engineering talent.
Amazon is also deepening its broader AI push through its relationship with Anthropic. Earlier this week, Amazon announced plans to invest up to US$25 billion in Anthropic as part of a wider cloud partnership under which the AI company is expected to spend more than US$100 billion on Amazon cloud infrastructure over the next decade.
That backdrop makes the Meta agreement more than a single customer win. It reinforces Amazon’s attempt to position AWS as a core AI infrastructure platform at a time when Nvidia remains the dominant provider of AI accelerators and other cloud providers are also promoting proprietary chips. Axios reported that Amazon will supply Meta with tens of millions of Graviton cores, with the possibility of even more in the future.
For Meta, the partnership adds another layer to its aggressive AI spending strategy. The company has been pouring capital into data centres, chips and model development as it seeks to embed AI deeper into Facebook, Instagram, WhatsApp and its advertising systems. The strategy is designed to defend Meta’s core business while positioning the company for the next phase of consumer and enterprise AI adoption.
However, the scale of AI investment has also raised questions among investors. Technology companies are spending heavily before the full monetisation of AI is visible. Meta, Amazon, Microsoft, Google and other major players are under pressure to show that their infrastructure spending can eventually translate into higher revenue, stronger productivity and defensible margins.
The deal also highlights a wider shift in the AI supply chain. The market is no longer only about securing Nvidia GPUs. Large technology companies are increasingly mixing different types of chips to optimise cost, performance and availability. CPUs, GPUs, custom accelerators and in house silicon are becoming part of a more diversified AI infrastructure stack.
For Amazon, Meta’s participation could strengthen confidence in Graviton as a serious commercial product. AWS already uses its custom chips across its cloud business, but major external adoption by a company of Meta’s scale sends a powerful signal to other enterprise customers that Amazon’s silicon strategy is gaining traction.
For the broader technology sector, the agreement shows how the AI boom is reshaping relationships between competitors. Meta and Amazon compete across advertising, digital services, commerce and consumer technology, yet they are also increasingly connected through cloud and infrastructure needs. In the AI era, rivalry and dependency can exist at the same time.
The Ledger Asia Insights
For Asian investors, Meta’s deal with Amazon should be viewed as part of a much larger infrastructure cycle rather than a narrow technology contract. AI is no longer only about software innovation. It is becoming an industrial scale investment theme involving chips, power, data centres, cloud platforms, networking equipment and advanced manufacturing.
This matters for Asia because the region sits at the centre of the global semiconductor and electronics supply chain. Taiwan, South Korea, Japan, Malaysia, Singapore and parts of Southeast Asia are all linked to different layers of chip production, testing, packaging, data centre development and cloud services. As US technology giants expand AI infrastructure spending, Asian suppliers could benefit through hardware demand, component orders, power solutions and data centre investment.
At the same time, the deal also carries a warning. The AI investment cycle is becoming more expensive and more concentrated. Only companies with deep balance sheets can afford the infrastructure needed to compete at the highest level. This could widen the gap between technology giants and smaller firms that must rent cloud capacity rather than build their own infrastructure.
For investors, the key issue is not simply whether AI demand is rising. It is whether the companies spending aggressively can turn that spending into durable earnings. Meta may benefit if AI improves advertising efficiency, user engagement and automation across its platforms. Amazon may benefit if AWS becomes a stronger infrastructure provider for AI workloads. But both companies still need to prove that higher capital expenditure can generate returns beyond market excitement.
The Meta Amazon deal is another sign that the AI race is entering a more mature and capital intensive phase. The winners may not only be the companies with the best models, but also those that can secure the most efficient infrastructure at global scale.
