Canada’s AI Strategy: The Ambition Needs Discipline

On June 4th, Prime Minister Mark Carney and his Team released AI for All, a multi-year, multi-billion-dollar national AI strategy. He framed it as Canada getting serious. For those of us watching where AI meets healthcare, the question isn’t whether Canada is finally moving; it’s whether we’ve learned anything from decades of standing still. 

AI for All isn’t Canada’s first AI strategy. It builds on more than four decades of federal AI leadership. From CIFAR’s pioneering AI research program in 1983, to the world’s first national AI Strategy in 2017, Bill C-27, the AI and Data Act (AIDA), that died on the order paper in January 2025.

What’s different this time is speed and scale. A one-month sprint last October delivered over 11,000 public submissions, 32 expert task force reports, and now, $2.3 billion in new federal spending. The Carney government seems to have learned Trudeau-era lessons: less legislative rigidity, more agility to match AI’s pace, and policy urgency on Sovereign AI, the geopolitical issue of the decade.

Now, Canada must transform its legacy funding vehicles managed by ISED into rapid reviews, fast announcements, and accountable programs that drive commercialization. This is critical in healthcare, where the speed of science and pace of tech adoption move in weeks and months, not the year or two it takes to stand up programs, run competitions, evaluate applicants, and finally move federal dollars to recipients.

But first, credit where credit is due. AI for All introduces something genuinely new to the Canadian playbook: the AI Missions Program, a “mini-moonshot” model backed by an initial $200-million flagship health mission, explicitly designed to push mature, validated clinical AI into frontline care. 

The strategy also expands the Canada CIFAR AI Chairs program from 143 to 200 researchers, commits $130 million to commercialization at AMII, Mila, and the Vector Institute, and AI hubs emerging in other provinces should align – not compete – and benefit. 

VITAL, with $210 million in combined federal, provincial, and institutional funding, is the AI Missions Program’s first announcement. It connects more than 160 hospitals across Ontario, Alberta, and Quebec (out of 1,087 hospitals nationwide) into a near-real-time health data network. That’s real infrastructure. VITAL extends the proven GEMINI platform developed at Unity Health Toronto; national expansion was the logical next step.

But we also have history. 

Canada Health Infoway has received more than $2.4 billion in federal funding since its creation in 2001. Yet, Canada still lacks the interoperable pan-Canadian electronic health record that lay at the heart of its original mission. Meanwhile, its flagship e-prescribing initiative, PrescribeIT, was shut down in 2026 after consuming nearly $300 million in public funding while processing fewer than 5% of Canadian prescriptions, prompting parliamentary investigations, governance reviews, scrutiny of executive compensation, and calls for greater accountability. 

Both the Prime Minister and AI Minister Evan Solomon were clear that TRUST is the currency of AI for All, so we cannot simply add more zeros to the same funding model that produced Infoway’s outcome. Strong governance, accountable KPIs that measure health system adoption and private-sector commercialization, not just publications, training and mentoring personnel, and academic excellence are non-negotiable.

Canada doesn’t have a research problem when it comes to foundational science. We have an adoption problem, and adoption problems don’t get solved by endowed Chairs and grants alone; they get solved by tight governance, competitive and timely processes, and genuine accountability from the industry players, not-for-profits, and academic institutions that receive and spend taxpayers’ money.

Low adoption isn’t an abstract risk in Canada; it’s a documented pattern. Canada’s decentralized Research Ethics Board system has long required multi-site studies to navigate multiple institutional ethics processes and local approvals, slowing clinical research. The problem became significant enough that CanReview was created to establish a single, pan-Canadian ethics review system for multi-site trials facing the same deployment challenges cited below. 

The strategy’s emphasis on AI literacy by training a million entry-level students and equipping 3,000 educators is laudable, but literacy without a system built with receptor capacity to absorb what people learn merely produces a more educated audience for the next stalled pilot. 

And here’s what the strategy doesn’t say aloud, and it must be called out; it’s the structure of healthcare itself. Canada isn’t building one AI-for-health system. It’s asking ten provinces, three territories, and a federal government with no constitutional authority over most of what we know as healthcare delivery to converge, cooperate, and deliver in real-time. 

Every AI Mission, sovereign health data platform, and clinical deployment must be negotiated with provinces, integrate vendors, and comply with varying privacy regimes. Scale that friction to 42 million people across 14 jurisdictions, and Canada’s own health-AI ambitions could end up stuck in the ICU before reaching a ward bed and successful discharge. We need an explicit federal-provincial-territorial governance instrument built for interoperability from day one.

Then there’s also our unicorn obsession disorder. Canadian innovation policy, in health and other fields, has a longstanding ambition to create flagship companies — the next BioChem Pharma, AbCellera, InSilico Medicine, or Anthropic — while not supporting enough our SME layer (combining federal, provincial, NFP, and private venture capital funding into a P3+ model).

It’s time to accept the obvious: we won’t outspend American, European, and Asian biopharma, MedTech, and AI at their own scale game. Chasing this fantasy is folly.

We must play a different game on a different board with different rules. Our scientific depth, research infrastructure, and moderate industrial scale and speed are perfectly suited to drive innovation in blood cancers, rare and ultra-rare diseases, and antimicrobial resistance; therapeutic targets often ignored by big biopharma.

This isn’t a consolation prize. This is choosing our race where we can sprint and win. Minister Solomon must explicitly define this vision rather than let commercialization funding drift toward the same handful of flagship bets that have defined — and limited — Canadian biomedical, health, and life sciences policy for a generation.

None of this is an argument against the strategy. Au contraire. It’s an argument for executing it differently, properly, with purpose and achievable outcomes. 

Canada has, for the first time, assembled the ingredients: dedicated funding that still needs to be leveraged, a mission-based model, expanded research capacity, and a federal government bravely staking its political capital on delivery. What we haven’t demonstrated is the discipline to convert those ingredients into adopted systems, accountable spending, and a commercialization strategy that doesn’t require us to pretend we’re building the next Silicon Valley. It just needs to work and be sovereign for us.

Success won’t be measured in press releases, photo ops, or Chair counts. It’s hospitals that deploy AI tools, healthcare professionals who use them daily, clinical trials infrastructure and processes that become the envy of the world, patients with better outcomes, and taxpayers who don’t wake up in five years to another Infoway-esque headline. 

Canada has never lacked ambition. We’ve lacked the discipline to land it.  This is the strategy that finally must prove the difference.  We are All in on AI for All.