AWS to raise prices for Nvidia-powered GPU instances by about 20% starting July 1

Amazon Web Services will raise the hourly prices it charges for multiple Nvidia-powered GPU instance families by roughly 20% starting July 1, according to a report published by Yahoo Finance. The change is aimed at customers using AWS accelerators built around Nvidia GPUs, and it covers several of AWS’s most powerful Nvidia-based instance lines.

The reported updates include a new hourly rate of $14.04 for the P6-B300 accelerator instance. AWS will also bill the P6-B200 at $12.355 per hour, under the pricing schedule that begins July 1.

For the P5 accelerator family, the report says AWS regions outside the United States will see a new hourly rate of $4.72, while U.S. regions will be priced at $5.191 per hour. Taken together, the listed figures reflect the broader 20% increase described in the report.

The reported rate table suggests AWS is adjusting pricing across performance tiers, from the P6 line at the high end to the P5 line at a lower hourly cost. For customers running workloads such as AI training, inference, and other GPU-accelerated tasks, the effective cost of compute time would increase proportionally if they keep the same instance types and run durations.

Nvidia supplies the underlying GPU hardware used in many cloud AI deployments. While AWS controls its own service pricing, changes in the cost and availability of cloud GPU capacity can influence how enterprises budget for AI projects and how frequently they run models in production or during experimentation. Traders and investors often watch GPU compute pricing for indicates about demand and supply conditions in the AI infrastructure buildout.

Cloud GPU pricing also tends to be one of the few levers companies can move quickly as hardware costs, configuration mix, and utilization change. Even when GPU demand stays strong, providers may adjust pricing to manage capacity, encourage specific workloads, or reflect new procurement and operating cost assumptions. The Yahoo Finance report, however, does not provide details on the drivers behind the AWS rate change.

The company did not outline, in the information provided in the report, whether the price adjustment is tied to specific hardware revisions, contractual procurement costs, or changes in how customers are billed for utilization. It also does not explain whether AWS expects to offer any corresponding discounts, reserved capacity changes, or new on-demand tiers to offset higher list pricing.

For Nvidia’s ecosystem, what matters next is whether GPU instance price increases are broad across regions and tiers, and whether they persist beyond the July 1 effective date. Companies using AWS for AI workloads will also be looking for guidance on how existing commitments, scheduling, and billing behavior will interact with the updated hourly rates.