According to a report by The Information, AWS managers told engineers during a meeting in May to reduce compute usage wherever possible, including on CPU-powered servers.
- The report said AWS has directed teams to cut compute usage by decommissioning idle EC2 instances and reallocating that capacity to customers.
- One engineer told The Information that it now takes several days to obtain CPU server capacity that previously became available within a few hours.
- AWS disputed that anything has changed, saying it continues to meet the overwhelming majority of compute needs while promoting efficient EC2 usage.
The artificial intelligence infrastructure crunch is no longer limited to memory chips and GPUs. AI demand is now straining CPU capacity as well, with a report saying Amazon.com Inc.’s (AMZN) AWS has asked engineers to conserve compute resources to ensure sufficient capacity for customers.
According to a report by The Information, AWS managers told engineers during a meeting in May to reduce compute usage wherever possible, including on CPU-powered servers. The report also said some engineers have experienced longer wait times to access CPU server capacity than in the past.
Amazon shares were up more than 1% in Friday’s opening trade.
AWS Reportedly Tightens Internal Compute Usage
The report said AWS has given teams deadlines later this year to reduce compute usage, including by decommissioning idle Elastic Compute Cloud (EC2) virtual server instances that are no longer needed and reallocating that capacity to customers.
One engineer told The Information that it now takes several days to obtain CPU server capacity that previously became available within a few hours.
AWS disputed that anything has changed in how it manages compute resources, saying it continues to satisfy “the overwhelming majority” of compute needs for both internal teams and customers while encouraging efficient use of EC2 resources.
CPU Demand Rises Alongside AI Boom
The demand for CPUs is rising as companies deploy more AI applications and agents, while CPUs also remain essential for AI tasks such as preparing data for model training.
Advanced Micro Devices Inc. (AMD) CEO Lisa Su said earlier this week that the AI market has reached an “inflection in demand,” particularly for CPUs, as enterprises deploy more AI agents and move experimental AI workloads into production.
She added that agentic AI is expected to increase, rather than reduce, the importance of CPUs, with CPU-to-GPU ratios moving to “better than 1:1” over the next three to five years.
Arm Holdings (ARM) CEO Rene Haas said last week that the industry’s largest AI infrastructure providers are increasingly adopting Arm-based CPUs alongside AI accelerators.
He pointed to Nvidia’s Grace and upcoming Vera processors, as well as Google’s Axion chips used with its in-house TPUs, as examples of CPUs playing a growing role in AI infrastructure.
Intel executives have also pointed to rising CPU demand from AI workloads. CEO Lip-Bu Tan said during a post-earnings call last month that demand tied to agentic AI and inference remains strong, while CFO David Zinsner said the ratio of CPUs to GPUs is approaching parity and could eventually tilt further toward CPUs on a unit basis as customers continue ramping AI infrastructure investments.
What Retail Traders Think Of AMZN Stock
Retail sentiment on Stocktwits around Amazon trended in the ‘bullish’ territory at the time of writing.
AMZN stock is up 20% year-to-date and 24% over the past 12 months. The S&P 500 ETF (SPY) is up 22% over the past 12 months, while the Invesco QQQ Trust (QQQ) is up 27%.
The iShares Semiconductor ETF (SOXX) is up 124% during this period, while the Global X Artificial Intelligence & Technology ETF (AIQ) is up 41%
For updates and corrections, email newsroom[at]stocktwits[dot]com.<