The sovereign imperative: Why enterprise AI mandates a new infrastructure playbook
150+ MW. By 2032, TELUS will operate a distributed sovereign AI grid across Canada with 60,000 NVIDIA GPUs — and it's already sold out.
Why it matters
Enterprise AI infrastructure is shifting from multi-tenant public cloud to dedicated, sovereign data centers. TELUS's Canadian AI factory deployment shows how regulated enterprises are architecting for data residency, compliance, and computational density — with real early-adopter validation from OpenText, Accenture, and League.
The key facts
8 to knowTELUS Sovereign AI Factory (Rimouski, QC) operational: 22.74 petaflops, ranked TOP500 #78
Architecture: HPE AI Factory + NVIDIA H200 Tensor Core GPUs (141GB HBM3e, 4.8 TB/s bandwidth per GPU) + NVIDIA Quantum-2 InfiniBand networking
Energy: 99% renewable hydroelectricity, 98% free cooling annually, 75% lower water consumption, 3x energy efficiency vs. standard enterprise data centers
Capacity sold out within months of commercial availability
Expansion roadmap: Kamloops, BC (Fall 2026); Vancouver M3 and 150 W. Georgia (late 2026–2029); total target 150+ MW, 60,000+ GPUs by 2032
Early deployments: OpenText (1,600 Canadian clients), Accenture (regulated verticals), League (healthcare), Mila (research)
Co-designed with Government of Canada's Enabling Large-Scale Sovereign AI Data Centres initiative
Inter-node latency elimination via unified HPC stack integration; NVIDIA Cloud Partner validation of NVIDIA AI Enterprise and NIM microservices
Go to the source
CIOcio.com
Publisher excerpt: For enterprise Chief Information Officers, the honeymoon phase of artificial intelligence is over. As organizations move past baseline experimentation and begin anchoring generative AI and agentic workflows into core operational stacks, we are hitting a collective wall. That wall isn’t defined by a…