Amazon triples Nvidia chip order amid surging AI demand
Tech giants expand partnership in the AI infrastructure race
1Order details
Amazon has tripled its Nvidia GPU order by adding 2 million more chips, including the new Blackwell Ultra, Rubin, and Rubin Ultra models. These GPUs will be delivered to AWS data centers in 2027 and 2028. This deal represents a significant increase from Amazon's 1 million GPU order made just five months ago. Nvidia stated that demand has exceeded their initial expectations, prompting this expanded agreement between the two tech giants.
2Deal magnitude
While financial terms weren't disclosed, based on GPU unit costs, the deal is estimated to be worth tens of billions of dollars. Nvidia reported $96.2 billion in Q2 revenue, with $89 billion coming from data center sales. The company forecasts $108 billion revenue for Q3, partly from next-gen Rubin GPUs. Nvidia has committed $279 billion to secure supply chain and manufacturing capacity for current and future data center projects, nearly triple last quarter's commitment.
3Technology integration
The partnership extends beyond GPU sales. Nvidia's networking hardware, open models, CPUs, data processing software, and robotics platform will be integrated across AWS. Nvidia CFO Colette Kress revealed Amazon will use Nvidia's full physical AI stack (Omniverse, Cosmos, Isaac, and Jetson) for its warehouse robots. AWS will also offer Nvidia's Nemotron family of open models through Amazon Bedrock and SageMaker, expanding AI service options for enterprise customers.
4Amazon's chip strategy
Despite this massive Nvidia order, Amazon continues developing its own AI chips. AWS is negotiating to sell its Trainium chips - direct competitors to Nvidia's H100 and Blackwell GPUs for deep learning - to other companies. Amazon's Arm-based Graviton CPUs also challenge traditional server chips from Intel and AMD. Amazon's custom chip business has reportedly surpassed $25 billion in annualized revenue, fueled by commitments from AI labs like Anthropic and OpenAI.
5Nvidia's expansion plans
Nvidia CEO Jensen Huang claims the company's Vera CPUs address a $200 billion new market opportunity. Nvidia's CFO stated Vera will be deployed by all major hyperscalers, AI labs, and system OEMs, with shipments already underway to lead partners like Oracle and SpaceXAI. Industry watchers are monitoring Rubin GPU sales in Q3 as an indicator of sustained demand for next-gen hardware. Nvidia began production shipments of Rubin this quarter.
6Industry demand surge
Both companies cited surging demand from startups, enterprises, AI labs, and governments as driving this expanded partnership. Nvidia's Huang noted AI is now generating profitable outputs, with more compute directly translating to more profit. As AI companies pour hundreds of billions into infrastructure, analysts are watching whether additional compute capacity will linearly increase profitability in this rapidly evolving market.
7Competitive dynamics
The partnership is notable as Amazon continues investing in competing chip technologies while making this massive Nvidia order. This reflects cloud providers' dual strategy of diversifying supply chains while reducing technological dependencies. Amazon's Trainium and Graviton chips could eventually challenge Nvidia's dominance, though Nvidia currently remains the undisputed leader in AI accelerator chips with its comprehensive hardware and software stack.
8Sector-wide implications
This mega-deal underscores the intensifying race for AI infrastructure. Cloud providers and tech giants are making massive investments to secure computing power for AI workloads. Nvidia's $279 billion commitment for supply chain and manufacturing capacity reflects industry growth expectations. These developments will shape the future of AI hardware and cloud computing markets, with implications across multiple sectors from robotics to enterprise software and beyond.
Original article: TechCrunch — Amazon, Nvidia çipleri için talebi üçe katladı