AWS raised prices on some Amazon EC2 Capacity Blocks for ML in January 2026, with reported increases of about 15% for P5e.48xlarge and P5en.48xlarge instances fitted with eight NVIDIA H200 GPUs. The change does not establish a price increase for all AWS GPU instances: Capacity Blocks use dynamic pricing, while AWS said its commitment on fixed On-Demand and Savings Plans pricing remained unchanged.
Which AWS GPU prices changed?
The January 2026 reports concerned EC2 Capacity Blocks for ML, a dynamic pricing option, rather than a blanket increase across AWS GPU offerings. The reported affected instance types were P5e.48xlarge and P5en.48xlarge, each described as having eight NVIDIA H200 GPUs. AWS’s instance guide history records P5e’s introduction on September 9, 2024, and P5en’s on December 2, 2024. (AWS EC2 Instance Types Guide history.)
Network World characterized the increase as approximately 15%. That is a useful description of the reported Capacity Block adjustment, not a rate that should be applied to every GPU instance, AWS region, or purchasing model.
What prices did reports cite?
The publications gave different hourly examples. They should be treated as separate reported examples, not combined into a single AWS-wide rate: region and product details can affect the price.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitches#1 Best Overall
- Standard Memory: 40 GB
- Host Interface: PCI Express 4.0
- Cooler Type: Passive Cooler
- Product Type: Graphics Card
| Publication and date | Reported example | How to interpret it |
|---|---|---|
| Network World, January 7, 2026 | P5e.48xlarge: $34.61 to $39.80 per hour | One published example; not a universal rate. |
| DataCenterDynamics, January 6, 2026 | US West example: $43.26 to $49.75 | A region-specific example; do not substitute it for the Network World figure. |
The reports said AWS had previously indicated that Capacity Block prices would be updated in January 2026, but that notice did not say whether rates would rise or fall. The exact effective timestamp and a complete region-by-region price schedule are not established by these reports.
Why did AWS say the price changed?
DataCenterDynamics quoted an AWS spokesperson explaining that EC2 Capacity Blocks for ML are dynamically priced according to supply and demand, and that the adjustment reflected the supply-and-demand patterns AWS expected for the quarter. The spokesperson also said AWS’s commitment not to raise prices on fixed pricing models such as On-Demand and Savings Plans remained unchanged. These are statements attributed to the spokesperson by the publication.
Rank #2
- GPU processor: NVIDIA RTX A5500
- CUDA cores: 10240
- 24GB GDDR6 ECC Graphics Memory
- System Interface: PCI-Express 4.0 x16
- 1 x DisplayPort to HDMI adapter
DataCenterDynamics also quoted Gartner VP analyst Lydia Long saying that AWS had always described Capacity Blocks as dynamically priced, comparing their fluctuations with Spot Pricing. Her comment is an analyst’s characterization, not an AWS statement.
What this means when comparing AWS GPU options
Do not compare an advertised Capacity Block rate with another EC2 rate without checking that the options match. The pricing model, region, instance type and GPU generation all matter. In particular, a Capacity Block figure does not by itself establish what an On-Demand or Savings Plans buyer pays.
Free tools Windows power users keep installed
One-click scans. No signup required.
- Check the exact instance type, including whether it is P5e.48xlarge or P5en.48xlarge.
- Match the AWS region before comparing hourly amounts.
- Confirm whether the displayed price is for a Capacity Block or a fixed pricing model.
- Verify the current rate directly for the region and purchasing option you intend to use; the published examples are not a complete regional schedule.
How this fits with other AWS price announcements
In June 2025, AWS announced price reductions of up to 45% for EC2 NVIDIA GPU-accelerated P4 and P5 instances. That announcement addressed a different pricing mechanism and does not directly reverse or cancel the later Capacity Block adjustment. (AWS News Blog, June 5, 2025.)
AWS separately announced that ECS Managed Instances GPU management fees would be reduced from July 1, 2026: 35% for G-series and 60% for P-series and Trainium. Those are managed-service fee changes, not evidence about EC2 Capacity Block rates. (AWS announcement, July 7, 2026.)
Quick Recap
Best Value
- Graphics Card Interface: Pci E
Rank #4
- Chipset: NVIDIA GeForce RTX 3090
- Video Memory: 24GB GDDR6X
- Memory Interface: 384-bit
- Output: DisplayPort x 3 (v1.4a) / HDMI 2.1 x 1
- Nvidia India 3 Year *
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
