Commission Vera Rubin, gate by gate.

Our Vera Rubin cluster in Western Canada is under construction. Current North American GPU operations use GB300.

The commissioning plan separates facility readiness, equipment integration, application qualification and operational handover. Each workstream has an owner, required evidence and open issues. Customer workload migration is planned against the current environment as well as the new platform.

Freeze the supplied system configuration and the OEM requirements for power, cooling, rack layout and service clearances. Match them to the actual site design.

Record the readiness checks and receiving plan before equipment arrives. A change in the system specification triggers a review of the bill of materials, facility interfaces and schedule.

Record drivers, runtime, communication libraries, framework, containers and application versions. Validate the delivered GPU topology, network and storage before application qualification.

Retain numerical checks, peak memory, communication tests and checkpoint/restart results. Compare the customer workload with its existing baseline under the same test inputs.

Commission access controls, monitoring, backup ownership, maintenance procedures and the support escalation route. Close outstanding test failures or document accepted exceptions.

The operational record names the configuration and responsible team. An installation milestone and a customer allocation are separate decisions; access begins after acceptance and commercial allocation are complete.

Inventory model artifacts, containers, datasets and software entitlements before migration. Test transfer, restart and output checks on the accepted platform.

Define the change window and fallback to the current environment. The readiness discussion establishes these requirements; the construction status remains visible until the commissioning gates and allocation are complete.