By Hamza in Quantum Weekly — Dec 22, 2025

Quantum #91

Issue #91 of the weekly HPC newsletter by HMx Labs. Things don’t appear to have slowed down for the holiday season at all, the news is mostly dominated by all things Slurm.

Last week was still running on all cores with no signs of letting up in HPC. The news was dominated by, rather unusually, by a fairly unglamorous (yet critical) bit of software that originates from a US national laboratory, Lawrence Livermore National Lab to be precise. I am talking, of course, about Slurm.

I’ve already written a little (see below) about its acquisition by Nvidia, but the other important Slurm related news was Google making releasing Cluster Director, a new Slurm based managed HPC offering to run on Google Cloud. This means that the three largest hyperscalers now all have a managed HPC offering based on Slurm with AWS Parallel Computing Service, Azure Cycle Cloud Workspace for Slurm and now Google’s Cluster Director. Just as Slurm itself becomes owned by one of their largest suppliers, Nvidia.

For the first time, in possibly a decade, the HPC scheduler market just got interesting again. I have more thoughts one this, but I will try and put finger to keyboard later in the week with another dedicated post.

And to cap it all off SchedMD (the company behind Slurm since 2010) also released a new version last week.

In The News

Updates from the big three clouds on all things HPC.

Next Platform provide a good history of Nvidia’s AI Enterprise stack in their story on the acquisition of Slurm but have no more answer about the future of Slurm than I did (see below for my take).