AWS Raises Nvidia-Powered GPU Rental Rates to $14.04 per Hour
NVDA•AWS will raise hourly rates for Nvidia-powered EC2 ML instances starting July 1, charging $14.04 for P6-B300, $12.36 for P6-B200 and up to $6.87 for P5en in U.S. regions, reflecting tight Nvidia-based compute demand. Industry discussions on creating futures markets underscore up to 4× price variance for identical Nvidia chips.
1. AWS Increases GPU Rental Pricing
Effective July 1, AWS will update EC2 ML capacity block rates for Nvidia-backed instances. New hourly charges include $14.04 for P6-B300, $12.355 for P6-B200, $5.191 for P5 and $6.865 for P5en in U.S. regions, marking the first adjustment since January.
2. Impact on Nvidia Compute Demand
The price increase indicates sustained tight supply and robust demand for Nvidia GPUs in cloud services. Higher AWS rates may boost cloud margins and support Nvidia's revenue growth while prompting customers to evaluate internal chips or alternative providers.
3. Proposal for AI Compute Futures Market
Industry participants are exploring futures or benchmarks for AI compute, aiming to financialize Nvidia-powered processing capacity. Technical complexities and heterogeneous infrastructure present significant obstacles to creating liquid, standardized compute contracts.
4. Price Variance and Benchmark Challenges
Identical Nvidia Hopper chips can command two to four times different rates across providers and locations, driven by factors such as data center reliability, power configuration and hardware stack optimizations. This variance complicates efforts to establish clear benchmarks.
Related News
Sources
