Amazon raises AI cloud prices amid rising memory costs

A second price increase this year for AWS's GPU reservation service reflects mounting pressure from AI hardware shortages, with businesses likely to face higher cloud computing costs that could eventually impact AI applications and enterprise software pricing.

Amazon Web Services (AWS) has announced another price increase for its EC2 Capacity Blocks for Machine Learning (ML) service, signalling rising cost pressures across the artificial intelligence infrastructure market. Effective from July, reservation prices for the GPU-based service will increase by around 20%, following an earlier hike of nearly 15% introduced in January.

EC2 Capacity Blocks for ML enables enterprises to reserve GPU capacity in advance to support AI model training and machine learning workloads. AWS said pricing for the service is reviewed periodically in line with changing supply and demand conditions but did not provide additional details on the latest revision.

The move comes at a time when demand for AI computing infrastructure continues to outpace available supply, forcing cloud providers to reassess pricing as the cost of critical hardware components rises.

Rising infrastructure costs may impact AI services

Unlike consumer price increases on smartphones or gaming consoles, AWS's latest revision directly affects the cloud infrastructure used by businesses to develop and deploy AI applications. As the world's largest cloud services provider, AWS supports millions of developers and enterprises that rely on its platform for AI models, enterprise software and digital services.

Industry observers believe the higher reservation costs could eventually filter through the AI ecosystem. Organisations with significant cloud computing requirements may face increased operational expenses, and some could pass those costs on to customers through higher subscription fees or software pricing.

The increase also mirrors a broader trend across the technology sector. Several major technology companies, including Apple and Xbox, have recently acknowledged rising component costs, while industry leaders have pointed to growing pressure from memory pricing as AI adoption accelerates.

High-bandwidth memory supply remains a key constraint

A major factor behind these cost increases is the limited availability of high-bandwidth memory (HBM), a critical component used in AI processors. Supply constraints have restricted GPU production, limiting how quickly cloud providers can expand AI infrastructure to meet surging enterprise demand.

Peter Berezin, Chief Economist at BCA Research, noted that constrained memory production ultimately limits GPU availability and data centre expansion. He added that the imbalance between demand and supply gives cloud providers greater flexibility to raise computing prices while maintaining strong customer demand.

The ongoing shortage has also benefited memory manufacturers such as Micron and SK Hynix, with investors expecting AI-driven demand to keep the memory market tight and infrastructure costs elevated for the foreseeable future.