Amazon Triples Nvidia GPU Order to 2 Million Chips

Amazon Deepens Nvidia Ties With Massive GPU Expansion

Amazon Web Services is dramatically scaling up its Nvidia chip purchases, agreeing to deploy an additional two million Nvidia GPUs across its data centers over the next several years 1. The move, described by Nvidia as a response to "surging demand" for AI computing power, represents roughly a tripling of the company's prior commitments and underscores how aggressively cloud providers are racing to secure chip supply for artificial intelligence workloads 1.

The expanded deal was formally disclosed by Nvidia executives during the company's latest earnings call. Nvidia CFO Colette Kress told analysts that AWS "is deploying an additional two million GPUs starting this quarter through the second half of fiscal year 2029," signaling a multi-year rollout rather than a one-time bulk purchase 4. Kress also indicated the partnership extends beyond raw chip volume, with Amazon set to adopt Nvidia's broader "physical AI" stack — software and infrastructure tools designed to support robotics and other real-world AI applications alongside traditional cloud computing 4.

More Than Just a Chip Order

Both Nvidia and outside coverage emphasize that this arrangement is not simply about Amazon buying more silicon. It reflects a deepening technical alliance in which AWS integrates Nvidia's full ecosystem — chips, networking, and software — more tightly into its infrastructure 1. This positions Amazon to compete more directly with rivals like Microsoft Azure and Google Cloud, all of which have been locked in an escalating buildout of AI-specific data center capacity to meet enterprise and consumer demand for generative AI services.

Nvidia's Broader Product Push

The AWS announcement lands amid a wave of Nvidia product news spanning both enterprise and consumer markets. On the high end, Nvidia's GB300-powered DGX Station — a desktop-sized AI workstation built around a 72-core Grace CPU and Blackwell Ultra GPU with 748GB of unified memory — has begun shipping to individual buyers starting at roughly $94,930 and reaching over $108,000 when fully specced 3. Though aimed at well-funded developers and enterprises rather than typical consumers, its availability illustrates how Nvidia is packaging data-center-grade AI power into increasingly accessible, if still costly, form factors.

On the gaming side, Nvidia has rolled out DLSS 4.5 in early access, bringing an improved AI-based Ray Reconstruction denoiser along with upgrades to Super Resolution and frame generation to RTX graphics cards 2. Separate consumer-focused coverage has also compared Nvidia's GPU lineup against gaming console hardware, such as identifying which Nvidia card roughly matches the PlayStation 5's graphics capability 5.

Why It Matters

Taken together, the AWS expansion and Nvidia's simultaneous consumer and enterprise product moves highlight the company's dual strategy: capturing massive infrastructure spending from hyperscale cloud providers while maintaining momentum in gaming and prosumer hardware. The scale of Amazon's order in particular signals continued confidence that demand for AI inference and training capacity will keep climbing, even as questions persist industry-wide about the long-term costs and returns of such enormous hardware investments.