Cloud provider offering ARM-based servers that cost less for the same work

AWS is a strong starting point: its ARM-based Graviton EC2 instances are often cheaper for the same workload—AWS says they can deliver up to 40% better price-performance than comparable x86 instances. (d1.awsstatic.com)

Other providers offer ARM options too: Google Cloud Tau T2A, Microsoft Azure Cobalt, and Oracle Cloud Ampere. (developer.arm.com)

The savings depend on your workload and software compatibility, so benchmark a representative workload before switching. Are you looking for a low-cost VPS, or compute for an existing application?

The provider you're thinking of is almost certainly AWS, with its Graviton processors, though Google Cloud (Axion) and Microsoft Azure (Cobalt) also sell Arm servers.

What AWS claims:

• AWS's own product page says Graviton-based instances cost up to 20% less than comparable x86-based EC2 instances. AWS Graviton-based instances cost up to 20% less than comparable x86-based Amazon EC2 instances.
• The bigger claim is price-performance, meaning cost per unit of work. AWS says Graviton provides up to 40% better price-performance compared to equivalent x86-based instances for a broad range of workloads.
• Managed databases follow the same pattern. For RDS, Graviton4 instances reportedly deliver up to 40% better performance and up to 29% better price-performance than comparable Graviton3 instances.

Caveats:

• The "up to" figures are best-case numbers from AWS benchmarks. Your savings depend on your workload.
• Some third-party articles claim about 50% savings from Intel to Graviton. I'd treat that as an outlier, since the official figure is closer to 20% on price, with more from performance gains.
• Compatibility is the main constraint. The main requirement is that your software runs on ARM, which is true for most interpreted languages, containers, and managed services, and increasingly for compiled code.
• Graviton doesn't cover everything. One comparison notes that Graviton is confined to general-purpose, compute-optimized, memory-optimized, storage, and HPC families on Linux, so GPU, bare-metal, and Windows-licensed workloads may still need x86.
• Single-threaded workloads may favor x86. One source notes traditional x86 instances maintain advantages in scenarios requiring high single-core performance.

Practical approach: Benchmark your own workload on a Graviton instance versus your current x86 instance, and compare cost per unit of work rather than hourly price. Then check that your dependencies have ARM builds.

I didn't verify current pricing or benchmarks for Google Cloud Axion or Azure Cobalt in this search, so check their current documentation if you're comparing providers.

Cloud providers offering ARM-based servers generally provide a better price-performance ratio than traditional x86 (Intel/AMD) equivalents, often resulting in cost savings of 20–40% for comparable workloads.

The following major cloud providers offer ARM-based instances, typically powered by Ampere Altra processors or custom silicon:

Major Cloud Providers

• Amazon Web Services (AWS): AWS is the most prominent leader in this space with its custom Graviton processor line.
  • Offerings: Graviton2, Graviton3, and the latest Graviton4 instances.
  • Benefit: AWS explicitly markets these as providing up to 40% better price-performance than comparable x86 instances. They are widely supported across managed services like RDS, ElastiCache, and Lambda.
• Google Cloud (GCP): Offers Tau T2A virtual machines, which are powered by Ampere Altra processors.
  • Benefit: Optimized for scale-out, cloud-native workloads like web servers, containerized microservices, and media transcoding.
• Microsoft Azure: Provides VMs featuring Ampere Altra processors (e.g., Dpsv5 and Epsv5 series).
  • Benefit: These instances are designed for scale-out, cloud-native applications and often deliver up to 50% better price-performance compared to similar x86-based VMs.

Why ARM Costs Less

• Performance-per-Watt: ARM architecture is based on Reduced Instruction Set Computing (RISC), which is inherently more power-efficient. Lower energy consumption and reduced cooling requirements allow providers to pass savings on to the customer.
• Higher Density: Because ARM chips are more efficient, providers can pack more cores into the same physical rack space, improving overall data center economics.
• Simplified Design: The focus on efficient, simple instruction sets allows for "leaner" performance that excels at throughput-heavy tasks like web serving, container orchestration, and microservices.

Important Considerations Before Switching

While ARM instances are almost always cheaper, they are not always a "drop-in" replacement for every workload:

• Software Compatibility: Your application and its dependencies must be compatible with the ARM64 architecture. Most modern languages (Python, Java, Node.js, Go) and containerized applications are easily recompiled, but proprietary software or legacy binaries built exclusively for x86 may not run.
• Workload Type: ARM excels at horizontal scaling, web serving, and containerized microservices. If your workload depends on specific x86 features (like certain AVX-512 instructions) or legacy Windows-based applications, you may find x86 instances more suitable.
• Testing: Always perform a performance audit on your specific application. While the "sticker price" is lower, you should verify that your specific application maintains the necessary performance levels (such as latency) to ensure the overall value is truly better for your use case.

AWS is a strong example: its Amazon EC2 Graviton instances use Arm processors and AWS says they can cost up to 20% less than comparable x86 instances ¹. The right choice depends on your workload and whether your software supports Arm—benchmark the actual job, since lower hourly pricing doesn’t always mean lower total cost. ²

Other major providers also offer Arm-based VMs, including Google Cloud Axion and Azure Cobalt. ³