Amazon to deploy 2 million additional Nvidia GPUs as AI compute demand soars
Amazon Web Services will add 2 million Nvidia GPU chips to its data centers in 2027 and 2028, tripling an earlier commitment made five months ago. The expanded partnership also integrates Nvidia's networking, software, and robotics technology across AWS. The deal, valued in the tens of billions of dollars, comes as Amazon continues developing its own competing AI chips.
The commitment marks a rapid escalation from the roughly one million GPUs Amazon pledged just five months ago, with Nvidia citing demand that has outpaced initial projections. The new deployment spans three upcoming chip generations and is scheduled to arrive across 2027 and 2028, while the partnership also extends into Nvidia's networking hardware, open-source models, data processing tools, and robotics platforms for Amazon's warehouse operations.
Amazon continues to pursue its own silicon strategy in parallel, with its Trainium chips positioned as direct alternatives to Nvidia's offerings and its Graviton CPUs challenging Intel and AMD. The company's custom chip business recently crossed a $25 billion annualized revenue run rate, though Nvidia's dominance remains evident in its quarterly results, which showed $89 billion in data center revenue alone.
This deepening partnership could shape the economics of cloud computing and AI development for years. Businesses relying on AWS for AI workloads may benefit from greater capacity and potentially more competitive pricing, while the scale of the deal underscores how concentrated AI infrastructure has become around a single supplier. Amazon's parallel investment in its own chips may eventually give customers alternatives, but for now, Nvidia's position appears entrenched, which could influence costs and access for startups, researchers, and enterprises dependent on AI compute.