Arm moves into the heart of the cloud stack

Hyperscaler adoption of Arm-based compute is accelerating multi-architecture infrastructure. As AI workloads scale and cloud demand grows, providers face pressure to increase performance while controlling power consumption, cost, and datacenter footprint. Major hyperscalers now offer Arm-based options, including AWS Graviton, Google Cloud Axion, Azure Cobalt-based instances, and Oracle Cloud Infrastructure Ampere Arm processors. Arm-based platforms aim to improve performance while reducing power consumption and overall cost. Reported results include up to 65% better price-performance and up to 60% greater energy efficiency across databases, AI inference, and networking services. Production deployments show large gains, including Spotify performance improvements with lower compute costs, Pinterest infrastructure cost savings and carbon reductions, and Uber’s multi-architecture migration across thousands of microservices.

"Five years ago, most cloud workloads ran on a single CPU architecture by default. No more. Today every major hyperscaler offers Arm-based compute, so what began as an option is now a core part of modern cloud infrastructure."

"The reason is straightforward. As AI workloads scale and cloud demand continues to grow, providers are under pressure to deliver more performance while controlling power consumption, cost, and datacenter footprint. Meeting those demands is forcing a rethink of the hardware foundations of the cloud."

"Arm-based cloud instances have demonstrated up to 65% better price-performance and as much as 60% greater energy efficiency across various workloads, including databases, AI inference, and networking services."

"At Uber, engineers have taken the multi-architecture approach even further. The company is integrating Arm-based hosts alongside x86 infrastructure across thousands of microservices as part of a broader migration to the cloud. It is shooting for increased hardware flexibility, improved price-performance, and support sustainability objectives while continuing to scale globally."

Read at theregister