Amazon 
AMZN According to Seeking Alpha, the company is increasing prices for GPU rental capacity on AWS, a sign that demand for AI computing power remains tight across major cloud regions.
AWS said new EC2 capacity block pricing for machine learning will go into effect on July 1 and will be available for instances powered by multiple Nvidia (NVDA) GPUs. Hourly prices per accelerator include $14.04 for P6-B300, $12.355 for P6-B200, $5.191 for P5 in the US region, $5.97 for P5e, and $6.865 for P5en in the US region. AWS said all other prices remain unchanged.
According to the company, the reservation price for EC2 capacity blocks for ML will be updated regularly based on supply and demand. The last price increase occurred in January, showing that AI infrastructure capacity remains expensive and in high demand as companies compete to secure GPU access.
By the way, this price adjustment specifically applies to Amazon EC2 Capacity Blocks for ML, a short-term GPU reservation option where pricing is dynamic and reflects supply and demand. AWS said the change does not apply to fixed price models such as On-Demand or Savings plans for GPUs and other general-purpose workloads.
The next thing to watch is whether customers absorb the increase or move more workloads to cheaper internal chips or alternatives.
