Introducing Z4M: Storage-Optimized Compute for AI/ML and Parallel File Systems
Google, Friday, November 6th, 2026
Google made its storage-optimized Z4M machine series generally available as VMs and bare metal for AI/ML and parallel file systems.
Google Compute Engine's Z4M machine series is now generally available as VM and bare-metal instances, aimed at AI/ML checkpointing, distributed KV cache and parallel file systems.
Built on fifth-generation Intel Xeon Scalable processors and Titanium, Z4M offers up to 168,000 GiB of Local SSD, 192 vCPUs, 1,536 GiB of memory and 400 Gbps of Cloud RDMA networking.
Google says RDMA cuts data-access latency by up to 60% versus TCP, and Z4M delivers up to 15.6M random-read IOPS and 75,600 MiB/s sequential reads, up to 70% and 110% better than Z3.
Bare-metal instances suit custom hypervisors, licensing needs and isolated agent sandboxes, and GPUDirect and TPUDirect Storage help keep accelerators fed.