Canada laid out its national AI strategy on June 4, 2026, with six pillars covering everything from sovereign compute to workforce development. But the country's semiconductor council says a critical piece is missing: a plan for the silicon that actually makes AI run.

On Wednesday, Canada's Semiconductor Council (CSC), which represents nearly 60 domestic chip companies, submitted a position paper to the federal government titled Sovereign AI Needs Sovereign Silicon. The message is blunt: without a deliberate strategy to build and source more chips at home, Canada's ambitions for AI sovereignty will remain incomplete, no matter how many data centres it builds on home soil.

The council isn't asking for a new standalone funding program. Instead, it wants Ottawa to graft a seventh pillar onto the existing AI strategy—one that sets targets for Canadian hardware in publicly funded compute projects, creates on-ramps for chip firms in existing AI programs, and adds semiconductor-specific talent policies to the government's workforce plans.

At first glance, Canada's AI strategy looks comprehensive. Pillar three, "Building a foundation for Canadian sovereign AI," explicitly targets domestic compute infrastructure. The government has already signed a deal with TELUS in May to advance sovereign AI data centres, and it has rolled out programs to help Canadian firms access high-performance compute.

But the strategy treats sovereign compute as a data-centre problem—lots of real estate and cooling, essentially—without paying much attention to what goes inside those racks. The processors, accelerators, networking silicon, and memory chips that actually train and run AI models are overwhelmingly designed and manufactured outside Canada.

"The government position is correct, in terms of 'no sovereign AI without sovereign compute,'" CSC managing director Paul Slaby told BetaKit, which first reported the council's push. "But what we want to say is 'there's no solving compute without a domestic capability in semiconductors.'"

That distinction matters for anyone building or buying AI infrastructure in Canada. A data centre hosted in Toronto that's full of Nvidia GPUs and Intel CPUs still leaves the country dependent on foreign supply chains for the most valuable—and most strategically sensitive—part of the stack.

A pragmatic retreat from a bigger ask

The seventh-pillar proposal is actually a scaled-back version of what the industry originally wanted. For years, chip firms had pushed for a standalone national semiconductor strategy, similar to what the U.S. and Europe have enacted. But in May, AI and Digital Innovation Minister Evan Solomon told BetaKit that a separate chip strategy wasn't happening.

Instead, the government argues that existing AI-compute programs and broader industrial funding already cover semiconductor companies. Solomon has pointed to the Sovereign AI Compute Strategy, federal support for the FABrIC edge-components initiative, the Canadian Photonics Fabrication Centre spinout, and a $210 million Strategic Response Fund commitment involving IBM and other firms in Bromont, Quebec.

Those are real investments, but they aren't coordinated under a single plan with measurable goals. For example, there's no procurement preference for Canadian-made photonics or packaging, no requirement that publicly funded data centres include a certain percentage of domestically sourced hardware, and no clear path for a chip design startup to plug into the government's AI talent programs.

What this means for businesses and IT teams

For now, the practical impact is limited to policy circles and procurement conversations. But the absence of a semiconductor pillar could shape where Canada's AI infrastructure dollars flow over the next several years.

If you're an enterprise IT leader: The government's current approach means that when public money helps build new AI compute capacity, there's no incentive to choose Canadian silicon—even where domestic alternatives exist. Photonics components, edge processors, and secure packaging services from Canadian firms will have to compete on price and performance alone against global giants like TSMC and Nvidia, without any policy tailwind.

Large-scale data-centre buildouts could end up simply importing foreign GPU clusters, reinforcing existing supply dependencies. That may be fine for short-term costs, but it leaves Canadian firms exposed to the same chip shortages and geopolitical risks that shook the industry during the pandemic.

If you're a Canadian chip company: The CSC's five asks give you a potential toehold in the government's AI spending. They include a Canadian-content plan for sovereign-compute investments, hardware entry points in existing AI programs, and semiconductor-specific treatment in workforce policies. If the government adopts even a few of these, it could create demand for domestic design services, packaging, and niche manufacturing that currently has no clear home in federal strategy.

If you're a researcher or startup: Access to sovereign compute is great, but if you're pushing the boundaries on novel AI architectures, you may need hardware that doesn't simply replicate what's already in the cloud. A domestic chip capability could eventually mean more specialized accelerators, lower latency edge devices, and compute options tailored to Canadian priorities like climate modeling or health privacy—but only if the policy creates a market.

How we got here

Canada's AI strategy didn't start from zero. The country has been a global leader in AI research for decades, thanks in part to pioneers like Geoffrey Hinton and the Canadian Institute for Advanced Research. The new six-pillar strategy builds on years of investment in talent and compute access, plus more recent moves toward "digital sovereignty" as geopolitical tensions have risen.

But the semiconductor gap reflects a broader industrial reality. Canada never developed a large-scale chip manufacturing base. Instead, its strengths lie in photonics, compound semiconductors, advanced packaging (like the IBM facility in Bromont), and design services. Those are valuable niches, but they don't produce the volume or profit of a memory fab.

Past attempts to rally their support have had mixed results. The Strategic Innovation Fund has sprinkled money into a few chip projects, and the FABrIC initiative aims to build an edge-computing components ecosystem. But no single agency or minister has had a clear mandate to tie those threads into a coherent national plan.

That's why the CSC's position paper is less about asking for new money and more about asking for a seat at the table. With a seventh pillar, semiconductor firms could see their needs reflected in procurement rules, workforce development, and the specific targets that Ottawa sets for sovereign AI.

What should happen next

For Canadian tech leaders and IT buyers, the immediate step is to pay attention to the government's response to the CSC proposal. Deputy Minister of AI Mark Schaan has reportedly given the position paper a "very positive" initial reception. That could lead to formal consultations or even a revision to the AI strategy's implementation guidelines.

Here are specific things to watch and do:

  • Track procurement language. If the government starts talking about "Canadian content" for sovereign compute, that's a signal that hardware spending could shift. Businesses that plan to bid on public-sector AI contracts should factor in possible requirements for domestic components.
  • Engage with the consultation. The CSC is soliciting feedback from its members, but any company that builds or uses AI infrastructure can likely submit comments through Innovation, Science and Economic Development Canada (ISED). The more specific the asks—for example, recommending that X% of a data centre's budget be reserved for Canadian-made photonics—the more likely they are to stick.
  • Re-evaluate supply chains. Even if Ottawa doesn't mandate domestic hardware, the conversation around sovereignty may accelerate a broader shift. U.S. and EU policies are already pushing chip production back to friendly shores. Canadian firms that rely on foreign silicon should start mapping alternatives, especially for applications where security or latency matters most.
  • Watch the workforce angle. The CSC wants AI workforce policies to include semiconductor training. That could affect university programs, immigration pathways, and co-op placements. If you're hiring chip talent, you might benefit from new incentives—or face more competition for grads from publicly supported labs.

The biggest test is still ahead

The federal government hasn't said "no" to a seventh pillar; it hasn't said much at all yet. Solomon's earlier rejection of a standalone chip strategy doesn't necessarily rule out folding semiconductors into the AI plan. Doing so would cost nothing in new funding and would give the minister a way to show the industry he's listening without reversing his earlier stance.

The real test will be whether the government moves from broad language about sovereignty to concrete procurement rules and program criteria. Data centres are easy to announce. Changing what goes into them—and who gets paid to put it there—is harder.

For Canadian chip firms, this is the closest they've come to a dedicated national strategy in a decade. For everyone else, it's a reminder that "made in Canada" AI depends on more than just where the server racks sit.