AI demand is overwhelming cloud computing, and sources say AWS requires engineers to shut down idle servers to reduce resource waste

IT House News on August 7, according to a report by "The Information" today, Amazon AWS recently put forward a rather "simple" requirement for engineers to shut down cloud computers that are no longer in use. According to reports, Amazon's request mainly targets idle EC2 instances. As AI consumes computing resources at an ever-increasing rate, the world's largest cloud service providers are beginning to feel the pressure of a computing power shortage. Data shows that AWS's resource underutilization problem is very serious, with approximately 65% ​​of EC2 instances having an average CPU utilization of less than 20% during a 30-day measurement period. Amazon has upgraded the Compute Optimizer (IT Home Note: Compute Optimizer) to automatically analyze server CPU usage and IO data within 14 days, and mark virtual machines with peak CPU utilization less than 15% and very low network traffic. Meanwhile, AWS added 3.8 gigawatts of new power capacity over the past year, enough to power 2.8 million homes. The reason for Amazon’s current shortage of computing resources is very clear. Cloud computing instances equipped with GPUs are the core resources for training and running large language models. Today's AI craze has made the cloud computing market fiercely competitive.