Will Canada’s New Posture Meet AI’s New Frontier?

Canada is entering a new era of economic nationalism. The industrial strategy announced by Prime Minister Mark Carney on September 5 puts sovereignty at the heart of federal policy and reflects a new posture. In the words of The Logic’s Kevin Carmichael: “Canada’s boy scout years are over.”

As the VP of Strategic Initiatives at a Canadian AI cloud startup, I welcome this shift in stance. It’s my hope that Carney’s vision for sovereignty, first of all, balances protection with a bold commitment to innovation, and secondly, reflects a clear understanding of where that innovation is actually happening today as it concerns the AI economy—the sector now defining our economic future.

Canada’s Innovation Challenge

Both goals are challenging on their own, and even more so when pursued together. In terms of innovation more broadly, Canada’s record is mixed. We’ve invested heavily in various programs, but the results have often failed to produce globally competitive companies. Government support is risk-averse and often fragmented. Frustrated entrepreneurs head south, and our own VCs channel capital into US startups while successful companies are often acquired before they can scale at home.

What’s needed now is not just more investment, but an approach that accepts uncertainty, tolerates failure, and concentrates resources behind big, original ideas.

The Real AI Frontier: Infrastructure

In terms of AI innovation specifically, it has yet to be widely recognized that the most strategic breakthroughs aren’t just happening at the level of apps or even models anymore. They’re happening deeper in the infrastructure stack. That may seem counterintuitive since compute—the underlying hardware—has become increasingly commoditized.

But infrastructure has evolved to include the orchestration, scheduling, optimization, and deployment layers that sit on top of the hardware. This broader infrastructure stack is where performance, cost, usability, and sustainability are defined—and where real innovation is now happening. Whoever leads here shapes how AI is built, deployed, and commercialized.

In short, infrastructure is no longer just an enabler of innovation; it is a strategic and dynamic site of innovation in its own right. It’s why we are now seeing AI infrastructure companies making forays into software and delivery of AI (e.g., CoreWeave acquiring Weights & Biases).

Where Performance and Innovation Converge

Designing better infrastructure means solving hard problems across a layered stack, from raw hardware to orchestration to the emerging platform layer, where infrastructure becomes a programmable, intelligent interface. This is where compute is allocated, throughput is optimized, and complexity is abstracted away for the enterprise user.

It’s also where cost-per-token efficiency, sustainability features like real-time carbon tracking, and user-friendly deployment experiences are built. The strongest advances come not just from isolated improvements at a single layer, but from vertical integration across the stack—in other words, tightly coupling software platforms with the underlying hardware to eliminate inefficiencies and unlock new levels of performance and control.

Without the platform layer, even the most advanced hardware can’t deliver performance, efficiency, or usability. In short, the platform that unlocks the value of the hardware.

The Canadian Opportunity

These aren’t incremental upgrades—they’re foundational shifts. And while we’ve seen real innovation at the orchestration and developer-experience layers, many platforms still mask underlying inefficiencies, especially around cost, sustainability, and infrastructure control.

A small but growing group of companies is rethinking how infrastructure is designed, optimized, and delivered. From cost-per-token efficiency to real-time carbon tracking, this is where meaningful differentiation is still possible—and where Canadian innovators like Radium have a real edge.

But those companies often struggle to gain visibility, support, or early backing because the risk profile is higher. Our procurement and investment systems still default to established players with compliance track records and predictable returns, but with less appetite or the core capability to support the kind of innovation needed today. We’ve seen this pattern before, and it threatens to repeat just as the platform layer becomes the key interface through which AI infrastructure is monetized and operationalized.

Sovereignty Through Innovation

That leaves Canada with a choice. We can pursue sovereignty as protectionism, emphasizing domestic control while defaulting to familiar players and legacy models. This may feel safer, but it leads to stagnation. Canada’s telecoms have entered the AI race, but so far this is largely through partnerships at the application layers.

Or we can pursue sovereignty by backing the companies that are building differentiated infrastructure, accepting the uncertainty that comes with it, and aligning ownership with the kind of innovation that defines global competitiveness.

The question becomes: will Canada’s sovereignty be built through the innovation of its startups, or absorbed by the incumbents aligning with them? And if the telecoms are to lead, can they evolve beyond a culture built for stability and compliance to one that fosters real innovation?