TELUS Tests Arrcus Network Fabric for Sovereign AI Inferencing
Event summary
- TELUS is running a proof-of-concept with Arrcus to deploy its Inference Network Fabric (AINF) for sovereign AI inferencing across Canada.
- AINF aims to deliver secure, low-latency AI for mission-critical applications like public safety and emergency response.
- The solution integrates NVIDIA BlueField-3 DPUs and Spectrum-4 Ethernet switches for secure, high-performance inference.
- Arrcus claims AINF can reduce Time to First Token (TTFT) by 60% and end-to-end latency by 40%.
- The PoC is part of a broader shift toward distributed, policy-aware AI networking.
The big picture
The PoC reflects a broader industry shift from centralized AI to distributed inferencing, where networking becomes the control plane for performance, sovereignty, and cost. For national operators like TELUS, this transformation is critical to delivering mission-critical AI services while maintaining data residency and regulatory compliance. The integration with NVIDIA’s hardware underscores the growing importance of specialized networking fabrics in AI infrastructure.
What we're watching
- Sovereign AI Adoption
- Whether TELUS can scale AINF to meet Canada’s sovereign AI infrastructure needs across public safety and enterprise sectors.
- Network Performance
- The pace at which Arrcus can demonstrate measurable gains in latency and throughput in real-world deployments.
- Vendor Lock-In
- How Arrcus balances hardware-accelerated performance with silicon diversity to avoid vendor lock-in for TELUS.
Related topics