AWS raises GPU prices 15% on a Saturday, hopes you weren't paying attention (theregister.com)

Article summary

AWS has raised the prices of its EC2 Capacity Blocks for ML by approximately 15%, affecting customers who reserve specific GPU instances for machine learning workloads. The price increase was made without prior notice, with the company citing supply and demand patterns as the reason. This move has raised concerns among customers and competitors, who see it as a potential trend in cloud pricing. The increase may impact companies with Enterprise Discount Programs or other negotiated agreements, potentially leading to higher costs.

Main themes

• Cloud Pricing
• GPU Costs
• Machine Learning
• Supply and Demand
• Cloud Computing
• AWS Pricing Strategy

What commenters say

• The price increase may be a sign of a new trend in cloud pricing, where prices can go up as well as down, contrary to the common assumption that cloud prices only decrease over time.
• The lack of communication from AWS about the price increase is seen as disturbing and may erode trust among customers, who expect transparency in pricing changes.
• Some argue that owning a GPU can be cheaper than renting one from a cloud provider, especially for those with constant or high-demand workloads, while others counter that renting provides flexibility and scalability.
• The price increase may be driven by the growing demand for GPUs in machine learning and AI applications, leading to supply constraints and higher costs for cloud providers.
• There are concerns that the price increase will lead to a shift towards subscription-based services, where consumers and businesses rely on cloud providers for computing power and storage, rather than owning their own hardware.
• Others see the price increase as a natural result of the growing demand for cloud services and the need for providers to balance their costs with revenue, and that it may lead to innovation and better consumer hardware in the long run.