Want to Run on AWS Graviton Without Paying Extra Licensing?

  • YellowDog Blog
Want to Run on AWS Graviton Without Paying Extra Licensing?

Why IBM Spectrum Symphony Customers Are Looking at YellowDog

AWS Graviton processors have become an increasingly attractive option for organisations running large-scale compute workloads. Built on Arm architecture and designed by AWS, Graviton offers strong price–performance and energy efficiency — particularly appealing for bursty, cloud-native, and high-throughput workloads.

For IBM Spectrum Symphony customers, however, accessing Graviton on AWS comes with an important caveat: an additional, architecture-specific software subscription. Ouch!

This is leading many teams to ask a simple question:

Is there a way to access Graviton — and other hardware types — without paying extra licensing just to use the silicon?

The answer is YellowDog.

The Hidden Cost of Graviton for Symphony Customers

IBM recently made Spectrum Symphony available on AWS Graviton. While this expands architectural choice, it also introduces a separate annual Graviton entitlement, priced per vCPU.

At list price, this means:

  • Customers pay their existing Symphony licenses
  • Plus an additional annual charge for Graviton support
  • On top of standard AWS infrastructure costs

For small environments this may be manageable. But for many others, especially teams managing infrastructure in Financial Services, running hundreds or thousands of cores, that Graviton uplift quickly becomes a six- or seven-figure annual expense — before a single EC2 instance is even launched.

The result? Graviton’s infrastructure savings can be partially or entirely offset by software licensing.

YellowDog: Hardware Choice Without Hardware Tax

YellowDog takes a fundamentally different approach.

With YellowDog, customers can:

  • Access AWS Graviton immediately
  • Use x86, Arm, GPU, on-demand, spot and preemptible instances
  • Run workloads across multiple clouds and regions
  • Without paying extra licensing based on CPU architecture

There is no “Graviton add-on”, no per-core uplift, and no need to renegotiate contracts just to try a different processor.

You choose the hardware that makes sense for your workload — not the one your scheduler happens to license most cheaply.

Why This Matters at Scale

For large or burst-heavy workloads, architectural freedom matters.

Many Symphony users:

  • Size licenses for peak capacity
  • Run workloads that spike for hours or days
  • Want to exploit spot markets and price volatility
  • Experiment with Graviton, GPUs, or newer instance families

Under traditional licensing models, each of these choices can trigger additional costs or restrictions.

YellowDog removes that friction.

Customers pay for:

  • The platform, not the processor
  • The compute they actually use, not theoretical peak cores
  • ​​Intelligent scheduling — not static capacity planning
  • Flexibility, not lock-in

YellowDog’s High-Throughput Scheduling (HTS) amplifies these benefits further. With processing speeds of up to 40,000 tasks per second, YellowDog enables faster scheduling, improving cost efficiency across Graviton and other CPU and GPU instance types as workloads grow.

We’re seeing clear evidence that Graviton is becoming central to scaled workloads on YellowDog. During peak demand, Graviton hours have grown by 3× month-on-month, while consistently maintaining their share during platform-scale events — a strong indicator that it scales alongside overall demand rather than being displaced by it. This usage shows a high correlation with spot capacity, reinforcing Graviton’s role in cost-optimised, elastic compute strategies. In practice, when customers scale up, Graviton scales with them, and in some cases now accounts for over 50% of a customer’s monthly compute usage, underscoring its transition from an alternative architecture to a primary choice for high-throughput workloads.

A Different Philosophy for Modern Compute

YellowDog was designed for a world where:

  • Hardware evolves faster than software contracts
  • Cloud capacity is elastic, not static
  • Cost optimisation depends on choice and speed, not long-term commitments

Rather than tying customers to a specific architecture or vendor, YellowDog lets teams:

  • Mix Graviton with x86
  • Burst when needed
  • Shift regions, instance types, or clouds
  • Optimise continuously — not annually

All without re-licensing their scheduler every time they adapt.

An Option Worth Considering for Symphony Users

For IBM Spectrum Symphony customers evaluating Graviton, the comparison is increasingly clear:

  • Symphony on Graviton
    → Additional per-vCPU annual licensing
    → Architecture-specific cost uplift
  • YellowDog
    → One platform
    → All hardware types
    → Superior 40,000 tasks per second throughput
    → No architecture tax

Graviton should be a cost-saving opportunity — not a licensing decision.

Learn More

If you’re currently running Symphony and exploring Graviton — or simply want more freedom to access Arm-based CPU instances alongside any other hardware across AWS, Google Cloud, and Azure — YellowDog offers a compelling alternative.

👉 Learn how YellowDog enables elastic, architecture-agnostic compute at scale
👉 Compare real-world costs for your workloads
👉 Start using Graviton without paying extra just to turn it on

Contact us to discuss using Graviton alongside all other CPU and GPU hardware options — because the hardware you choose should be driven by performance and price, not licensing constraints.

Author

Bruce Beckloff

Bruce Beckloff

CEO