Sovereign AI – Software
as Infrastructure

Sovereign AI: Where Sovereignty Actually Lives

“Sovereign AI” means at least five different things. Only some of them reduce your risk. Learn which ones matter and how to build them sustainably.

Not all sovereignty is created equal

To some, sovereignty means your data stays inside the border. To others, it means your models train at home, or a national company runs the facility. It can mean you own the building, or simply that you don’t depend on a foreign cloud. Each of these has a different cost, and some of them are procurement languages that change nothing. Canada can get this right by building sovereign AI infrastructure that’s efficient enough to be sustainable. Software is how you get there.

What you’ll learn

• Which layer of sovereignty carries the real risk: data residency, key custody, model weights, or who operates the facility
• Whether a sovereign cloud region is enough when the operator answers to a foreign parent company
• How software gets more performance out of every GPU and every watt
• Why security and tenant isolation have to be built into the architecture, not just written into policy
• Where sovereignty is theater, and what doing it right actually costs

Meet Our Speakers

Jordan Nanos

SemiAnalysis (moderator)

Val Bercovici

Chief AI Officer, WEKA

Dan van der Ster

CTO, Clyso

Ryan Grant

Associate Professor of Electrical and Computer Engineering, Queen’s University