Back in 2024 Google rolled out their Axion in-house ARM processors with the Google Cloud C4A instance type. Today they are expanding their Axion offerings in Google Cloud with the N4A instances now out of preview. The Google Cloud N4A instances are designed for scale-out web servers and microservices, containerized applications, back-end application services, databases, data analytics, and cost-effective development/staging/testing environments.
Google Cloud N4A instances have been in preview since the end of 2025 while today reached GA state. Compared to the C4A instances that allow up to 72 vCPUs and 576GB of RAM and 100 Gbps networking, the more cost-effective N4A instances are available up to 64 vCPUs, 512GB of RAM, and 50 Gbps networking. Like C4A, the N4A instances do have high memory (1:8 vCPU to memory) and high CPU (1:2 vCPU to memory) options to complement the standard 1:4 ratio for vCPU to memory size.
The Google Axion CPUs powering the N4A VMs remain on Arm Neoverse N3 cores and paired with Google’s Titanium IPU. N4A VMs operate within a single node for uniform memory access.
For getting an idea for the Google Axion performance with the new N4A instances, Google kindly provided some free access in advance of today’s GA date for some preview testing. To see the performance of Google N4A, using the 16 vCPU standard size I ran dozens of benchmarks comparing it to the prior-generation Tau T2A that is considered their prior-generation VM to the N4A series. Google Cloud T2A is powered by Ampere Altra processors based on Neoverse-N1 cores, now with N4A jumping ahead to Neoverse-N3.
Ubuntu 25.10 AArch64 was run on both the N4A and T2A 16 vCPU standard instances for looking at how the N4A performance is and how the ARM performance in Google Cloud has advanced since the T2A introduction back in 2022.
Pricing as of writing for the N4A-standard-16 size with 400GB of storage was $0.71 hourly or about $517.68 per month. The T2A-standard-16 with 400GB of storage was $0.64 hourly or $465.68 per month.
