🟢 🏠 SCALE comes home: NVIDIA GPU support is here. Compile your CUDA code to any target. Read the online documentation to learn more.

🟢 🏠 SCALE comes home: NVIDIA GPU support is here. Compile your CUDA code to any target. Read the online documentation to learn more.

Built for every layer of the accelerated compute stack

Whether you're bringing new hardware to market, running a GPU cloud, or operating mission-critical HPC workloads — we have a model designed for you.

Silicon Vendors

Support the CUDA toolkit from day zero

Join our design partnership program for early compiler access and joint go-to-market.

Learn more

Cloud Providers

Expand your GPU catalogue seamlessly

Offer CUDA-compatible compute on AMD, NVIDIA and alternative hardware with flexible GPU/h pricing.

Learn more

HPC Enterprise

Enterprise-grade support for mission-critical workloads

Custom agreements, dedicated engineering, and SLA-backed support for production deployments.

Learn more

Design Partnership Program

Launch with CUDA support from day one

New GPU architectures succeed or fail based on software support. SCALE reduces that risk by providing a production-ready CUDA compilation path from day one.

As a design partner, your hardware team works directly with Spectral engineers during the bring-up phase.

The result: your new silicon ships with CUDA support from day zero. Enterprise customers can evaluate existing CUDA workloads immediately.

  • Early access to SCALE compiler source and roadmap
  • Dedicated Spectral engineer embedded in your bring-up process
  • Joint reference workloads and benchmark suite
  • Co-marketing on launch and public roadmap mention

GPU/h Pricing for CSPs

Expand your GPU catalogue without new code

Your customers likely have existing CUDA workloads. They expect to run them with minimal friction. SCALE makes that promise real: AMD, NVIDIA, and alternative accelerators become first-class CUDA targets without requiring customers to rewrite a single line.

We offer a per-GPU-hour licensing model that scales linearly with your fleet — no upfront commitments, no per-seat overhead. Margins stay predictable as you grow.

Differentiate your cloud offering from hyperscalers who are locked into a single silicon vendor, and give customers the CUDA developer experience they deserve on the hardware they choose.

  • Flexible GPU/h licensing that scales with your fleet
  • Deployment support and integration documentation
  • Customer migration tooling for existing CUDA workloads
  • SLA-backed uptime commitments for the SCALE runtime

Custom Agreements & Dedicated Support

Enterprise-grade support for mission-critical workloads

SCALE for HPC enterprise gives you a named support contact, priority triage, and an SLA that matches the criticality of your deployments.

Licensing is negotiated to fit your organisation — whether that is an on-prem cluster, a hybrid cloud, or a research consortium with dozens of sites. We work with your procurement and legal teams to reach terms that make sense.

You keep your existing CUDA codebase and your existing validation suite. We make it run on the hardware your organization wants to provision or has already procured.

  • Named support contact with a dedicated communication channel
  • Custom licensing structures for on-prem and hybrid deployments
  • Prioritised bug triage and patch SLAs
  • Annual technical reviews and roadmap alignment sessions

Get in touch

We'll follow up within one business day.

Tell us about your hardware, workloads, or use case and we'll route you to the right person on our team. After submitting, you'll get a link to book a call directly.