HPC Cloud Updates WE 26 Oct 2025
Updates to AWS, Azure & GCP in the last week relevant for HPC practitioners. Google compile and run 100k applications across multiple CPU ISAs. AWS makes another tweak to PCS and gives us more EC2 metrics in CloudWatch and Azure has a new region coming up.
AWS
Need a hand picking EC2 instance types for your scientific HPC workload?

Predictive EC2 autoscaling now works in more places
Pick up on I/O limits being exceeded with new CloudWatch metrics
PCS can now rotate you secret keys for you
New Regional Instances: i8g in London, Singapore and Tokyo, c7i-flex in Jakarta
Azure
Azure now lets you turn off SMT and control how many cores are enabled in a VM. The latter is new but the former of those we’ve been doing for a number of years now and was always possible… At least via ARM templates or APIs though it did need a support request to enable it on the subscription
New region in Malaysia now available in private preview
Google Cloud
Google now compile and run 100,000 applications for both x86 and ARM CPUs. I’m actually genuinely impressed!

G4 VMs with Nvidia RTX PRO 6000 is now Generally Available

Google’s showing off their new H4D VMs at STAC Summit in NYC



