State of the art of arm compute - Q2 2025

Go back to list

State of the art of arm compute - Q2 2025
Compute Virtual-Machine Performance Price
Public

The cloud computing landscape has undergone a significant transformation with the emergence of ARM-based processors as a viable alternative to traditional x86 architectures. This shift represents more than just a technological evolution—it signals a fundamental change in how cloud providers approach performance, energy efficiency, and cost optimization. ARM processors, originally designed for mobile and embedded systems, have matured to deliver compelling performance characteristics for cloud workloads while offering potential advantages in power consumption and total cost of ownership.

This study provides a comprehensive comparative analysis of ARM-based Infrastructure as a Service (IaaS) offerings from four major cloud providers: Amazon Web Services (AWS), Microsoft Azure, Google Cloud, and Oracle Cloud Infrastructure. The research focuses specifically on instances configured with 8 virtual CPUs and 32GB of RAM, representing a common configuration for mid-tier production workloads across enterprise applications, web services, and data processing tasks.

Below you'll find the VM flavor used for this benchmark:

m8g.2xlarge
EC2 M8g powered by Graviton4, arm processors designed by AWS themselves and released in Q3 2024.
Standard_D8ps_v6
Dpsv6 powered by Azure Cobalt 100 released in Q2 2024.
c4a-standard-8 Axion
C4A VMs are powered by Google Axion processors released in Q2 2024
VM.Standard.A2.Flex.4-32

Ampere A2 powered by AmpereOne processors released in Q2 2024.