Government of Canada / Gouvernement du Canada in collaboration with
Invitation to pilot
Shared Services Canada · ICElab

Sovereign compute from infrastructure Canada already owns

Run your Monte Carlo simulations, astrophysics data analysis, microsimulation models, bioinformatics and genomics, AI batch inference, ETL pipelines, audiovisual processing, and parallel workloads of any kind

Securely distributed by DCP across infrastructure the Government of Canada already owns, for a fraction of the price of foreign hyperscalers.

Express interest

Elbows-up compute

Reroute workloads, and the spend that goes with them, to Government of Canada infrastructure instead of foreign data centres. Same card, same developer-friendly APIs, with Canadian data and dollars staying in Canada. Powered by the Distributive Compute Platform (DCP), a Canadian technology piloted by Shared Services Canada (ICElab, 2026).

Up to 4 to 50x lower cost compute

Same compute job, same results, for a fraction of the hyperscaler price. Data stays private and under sovereign control.

Already running on DCP

Monte Carlo simulation (DRDC), astrophysics data analysis (NRC-CanDIAPL), microsimulation and ensemble modelling, bioinformatics and genomics, batch inference, federated learning, ETL pipelines, audiovisual processing, and much more.

Independently security tested

The DCP Worker has been penetration tested by trusted external security firms. Security documentation is available under NDA.

Pilot terms

  • No onboarding cost
  • Low-effort onboarding: Distributive adapts the workload for you
  • Evaluate ROI before deciding to switch
  • No platform lock-in
  • Not subject to the US CLOUD Act
  • Proudly Canadian
Elbows-up compute for the Government of Canada

What the pilot entails

Four steps, over a few weeks. Your team brings a real workload; we show you what it does on DCP, and your department decides with the evidence in hand.

1

Meet SSC and Distributive

Your team, Brian McKittrick (SSC) and Dan Desjardins (Distributive)

A working session with your technical and security people to answer whatever you need answered: security and privacy questions, how data moves and where it lives, and what running on GC infrastructure looks like. We identify one or more suitable use cases, walk through the process below, and confirm next steps.

2

Send us a workload you run on AWS or Azure

Your team to Distributive

Under NDA, your team sends Distributive sample code, a representative data set and sample outputs for a compute workload your department currently runs on AWS, Azure or another commercial cloud. Distributive adapts it to run on DCP on GC infrastructure. Alternatively, Distributive can show your developers how to adapt their own code. The pilot measures four things:

  • aExactness. Parity of results against your current cloud output.
  • bPerformance. Run time compared to your cloud setup.
  • cCost. What the same run costs on DCP versus your cloud bill.
  • dPrivacy and sovereignty. Where your data and compute actually ran.
3

Run it and get the results

Distributive and SSC

We run the adapted workload on the SSC ICElab cluster and return the measured outcomes on all four points, along with the adapted source code and documentation of what changed and how to use it. Everything is yours to keep.

4

Decide with the evidence in hand

You

If the ROI is there, your department can move the workload onto DCP for real, pay-as-you-go with no platform lock-in. If it is not, you have lost nothing but a few hours of your team's time, and you have a hard number for what the alternative would cost.

GC data and dollars stay in Canada

Up to 4 to 50xlower cost than hyperscalers, depending on the workload
Hyperscalers DCP on GC infrastructure
Pricing model Reserve and rent capacity whether you use it or not Pay only for compute actually consumed
What you pay for Reserved instances, idle time, egress fees Computed results. Nothing else.
Cost visibility Billing surprises at month end Finite and bounded by the number of results
Data location Foreign-controlled data centres Infrastructure the Government of Canada already owns
Platform ownership US-headquartered corporations Canadian-owned and operated by Distributive
Data residency Data can leave the country Never leaves GC infrastructure
Workload isolation Shared infrastructure, unclear tenant boundaries Isolated sandbox per job
Commitment Reserved instances, switching costs, lock-in Pay-as-you-go: buy credits, run workloads, stop anytime
Jurisdiction Subject to foreign law including the US CLOUD Act Canadian law; not subject to the US CLOUD Act

The pilot runs your actual workload side by side on DCP and on your current cloud provider, so the comparison is on your numbers, not ours.

Express interest

Tell us about your workload and we'll be in touch. You can also reach out to Brian and Dan directly by email.

What the workload does, roughly how it is built (language, frameworks, GPU or CPU), how often it runs and how big a typical run is. More detail is better.

I am interested in
Live

Live cluster

Otherwise idle or underused GC infrastructure runs the DCP Worker, which pools its spare CPU and GPU capacity and puts it at your fingertips through the DCP Compute API. Departments that move off foreign hyperscalers pay a fraction of the price, and the spend stays within the Government of Canada.

–
DCP Workers
 
–
CPU cores
 
–
Compute value
Yearly value of unlocked compute capacity
Name Worker ID Last active CPU cores Slices running Earnings Earning rate
Loading live data…