Amazon and Nvidia announced on Wednesday an expanded partnership that will add 2 million Nvidia GPUs to Amazon Web Services (AWS) data centers. The update was disclosed during Nvidia’s quarterly earnings call.
The ordered chips — including Nvidia Blackwell Ultra, Rubin and Rubin Ultra GPUs — are built for the heavy compute required to train and run advanced AI models. Nvidia said these GPUs will be delivered to AWS in 2027 and 2028.
This expansion follows an agreement from five months earlier in which Amazon committed to deploy more than 1 million Nvidia GPUs across AWS starting this year. Nvidia said in a statement that since that announcement, “demand has exceeded those expectations.”
Neither company released financial terms for the new order, so the precise financial return for Nvidia is unknown. Based on typical GPU unit costs, the deal is likely worth tens of billions of dollars.
Broader integration beyond chip sales
The partnership goes beyond additional chip purchases. Nvidia said its broader technology stack — including the networking hardware that links thousands of GPUs into a single system, its open models, CPUs, data processing software and robotics platforms — will be integrated across AWS. The companies cited “surging demand” from startups, enterprises, AI labs and governments as a driver of the closer collaboration.
Amazon is simultaneously investing in its own chip programs
The expanded deal comes while Amazon continues to develop its own AI chips, particularly CPUs used as general-purpose processors in servers. Amazon has sought to reduce dependence on Nvidia and position its chips as competitors. Peter DeSantis, Amazon’s AI lead, has said AWS is in talks to sell its Trainium chips — positioned as a direct alternative to Nvidia H100 or Blackwell chips for deep learning workloads — to other companies for use in data centers.
Amazon’s Arm-based Graviton CPU is also viewed as a challenger to traditional server chips from Intel and AMD. On its latest earnings call Amazon said its custom chip business has crossed a $25 billion annualized revenue run rate, underpinned by $225 billion in total commitments from AI labs such as Anthropic and OpenAI.
Nvidia remains central to AI compute
Despite Amazon’s chip efforts, Nvidia remains a dominant supplier. Alongside the 2 million additional GPUs, Nvidia said it plans to ship an unspecified number of Vera CPUs to AWS — “some integrated with Rubin, others standalone,” according to Nvidia Chief Financial Officer Colette Kress.
Nvidia Chief Executive Officer Jensen Huang has touted ambitious expectations for Vera CPUs, saying in May he had identified a “brand new $200 billion TAM” for the company. Kress added that Nvidia expects Vera to be deployed by “every major hyperscaler, neocloud, AI lab, and system OEM,” with shipments already underway to lead partners including Oracle and SpaceX AI.
Extension into robotics and enterprise services
The partnership will also extend to Amazon’s warehouse robotics and enterprise offerings. Kress said Amazon plans to adopt Nvidia’s full physical AI stack to power its robot fleet: Omniverse (simulation and digital twin platform), Cosmos (world model platform), Isaac (robotics development platform) and Jetson (computing hardware for robots and edge AI). Nvidia also introduced a new version of Jetson this week aimed at making robotics computing more accessible for entry-level edge AI.
On the enterprise side, AWS will serve Nvidia’s Nemotron family of open models on Amazon Bedrock, its managed foundation-model platform, and on SageMaker, its managed machine-learning service.
Nvidia’s recent financials and manufacturing commitments
Nvidia reported $96.2 billion in sales for the quarter, beating analyst estimates. Data center revenue accounted for the bulk of sales at $89 billion, up 117% year over year. The company expects third-quarter revenue of $108 billion, some of which it says will come from next-generation Rubin GPUs; Nvidia said production shipments of Rubin began this quarter.
Nvidia has committed $279 billion to secure supply and manufacturing capacity for current and future data-center projects, a significant increase from $119 billion last quarter. That commitment includes $92 billion of projected spending for the remainder of the fiscal year and another $87 billion for fiscal 2028.
Management perspective and investor focus
Jensen Huang told callers that the critical industry development is that AI is now performing productive and useful work: “AI is generating profitable tokens…If we had more compute, we could generate more profitable tokens, which results in more profit for all of the services. This is the exact phase where we’re at, which is the reason why everybody’s leaning in.”
Investors will be watching whether the additional compute capacity translates into proportionate profits as AI companies invest hundreds of billions of dollars in infrastructure over the coming years.
