Canada’s ambitions in artificial intelligence and advanced scientific research are increasingly being shaped by a hard reality: progress now depends as much on domestic computing infrastructure as on talent or ideas. That shift was underscored this week with the deployment of a new supercomputing system at the University of Waterloo, designed to support thousands of researchers working across AI, health, climate science and engineering.
The system, known as Nibi, has been deployed within SHARCNET, Canada’s largest academic high-performance computing consortium, through a collaboration between Nokia and Hypertec. It will be available to more than 4,000 researchers each year across 19 partner institutions, with capacity open to academic researchers nationwide.
While the announcement focuses on a single installation, it reflects a broader recalibration underway in how countries think about AI research infrastructure. As models grow larger and simulations more complex, access to reliable, scalable compute has become a strategic concern, not just an operational one.
AI workloads push universities toward new infrastructure choices
Nibi represents Nokia’s first deployment of this class of AI-HPC data centre networking in North America. The system combines Nokia’s data centre fabric and IP networking with Hypertec’s system architecture and integration capabilities, including advanced immersion-cooling techniques designed to support dense, power-hungry workloads.
For SHARCNET, the move also marks a significant architectural decision. The consortium has shifted to an Ethernet-based interconnect, reflecting a wider trend in AI and HPC environments where flexibility, scalability and ecosystem compatibility are becoming as important as raw performance.
John Morton, director of technology at SHARCNET, said the combined solution delivered the scalability and reliability needed to support a wide range of demanding workloads. His comments point to a challenge faced by many academic institutions: AI research no longer fits neatly into traditional HPC patterns, and infrastructure must now accommodate highly variable, data-intensive tasks alongside conventional simulations.
At the University of Waterloo, which hosts the new system, the deployment is positioned as a material expansion of research capability. Charmaine Dean, vice-president for research at the university, said the collaboration provides access to world-class AI and high-performance computing infrastructure, reinforcing Waterloo’s long-standing role in computational research.
Sovereign compute becomes a strategic asset
Beyond the technical specifications, the project highlights Canada’s growing emphasis on sovereign AI infrastructure. By designing, deploying and operating advanced systems domestically, institutions reduce reliance on external cloud providers and retain greater control over data, workloads and research priorities.
Hypertec’s role as system architect and prime integrator reflects that intent. Mike Marracino, president of Hypertec Solutions Partner, said the project demonstrates Canada’s ability to deliver AI and HPC infrastructure to global standards, strengthening national competitiveness in research and AI development.
For Nokia, the deployment sits within a wider expansion of its Canadian footprint. The company recently broke ground on a next-generation campus in Ottawa’s Kanata North Tech Park, a nearly 750,000-square-foot facility focused on AI-powered networks, data centre networking, quantum-safe infrastructure and future 6G technologies. With more than 2,700 employees nationwide, Nokia is positioning Canada as a key hub for its research and development activity.
Jeff Maddox, president of Nokia Canada, described the Nibi project as an important step in maintaining Canada’s leadership in computer science and AI-driven innovation, signalling expectations of further collaborations of this kind.
From research capacity to national capability
The significance of Nibi lies not just in the number of cores or the speed of its interconnect, but in what it represents. AI research is increasingly constrained by access to infrastructure that can scale with ambition. Universities and public research bodies are being forced to think like operators of critical digital infrastructure, balancing performance, resilience and long-term sustainability.
By combining advanced networking, dense compute and energy-efficient cooling within a sovereign framework, the SHARCNET deployment offers a glimpse of how national research systems may evolve. As AI becomes central to fields ranging from medicine to climate modelling, the ability to run large-scale experiments at home, on trusted infrastructure, may prove as decisive as any individual breakthrough.
In that sense, Nibi is less a single supercomputer than a statement of intent: that in the AI era, research leadership increasingly rests on who controls the machines that make discovery possible.
