Amazon Web Services expands Nvidia GPU fleet by 2 million units, and Nvidia shares slide

Amazon Web Services is expanding its Nvidia GPU capacity more aggressively than previously outlined, according to a Yahoo Finance report published on Aug. 27.

The article says AWS is adding 2 million Nvidia GPUs, taking a planned build from 1 million processors to a total fleet size that exceeds 3 million. In practical terms, the update points to a faster scaling of the hardware used to run cloud-based AI and other accelerated workloads.

Despite the apparent increase in demand for Nvidia’s data center accelerators, the report also notes Nvidia shares fell on the news. The combination of “more GPUs” and “lower stock” suggests investors may be weighing other factors alongside AWS’s incremental build, such as near-term pricing, supply expectations, or broader market sentiment.

Nvidia’s investors have closely monitored large cloud customers because the company’s revenue profile depends heavily on data center demand, including orders and deployments tied to major AI infrastructure builds. Any change to the pace or size of those deployments can move expectations for what Nvidia can sell in coming quarters.

The Yahoo Finance report, as provided in the current packet, does not include additional specifics on the terms of AWS’s expansion, such as whether the incremental GPUs are tied to particular Nvidia product configurations, delivery timelines, or pricing. It also does not break out whether the “planned” and “expanded” figures refer to specific GPU generations, as opposed to GPUs broadly.

A further limitation is that the excerpted material does not confirm whether the revised fleet totals represent committed purchases, contracted capacity, or an internal deployment target. Without those details, it is difficult to translate the reported scale directly into a forecast for Nvidia’s revenue and margin trajectory.

Still, the magnitude of the reported ramp matters for the AI supply chain. Cloud providers are the core buyers of high-volume GPU systems, and large adjustments in fleet size can affect how quickly hyperscalers can train models, run inference at scale, and satisfy demand from enterprises and developers.

What to watch next is whether Nvidia or AWS provides additional disclosure around deliveries, contracted volumes, or supply plans. Nvidia typically reports quarterly results that reflect order and shipment momentum, while cloud customers may announcement infrastructure changes through announcements, procurement updates, or earnings commentary, all of which could help clarify the financial read-through of AWS’s larger-than-planned GPU buildup.