Canada’s AI for All strategy made a bet that most people skimmed past. Among the strategy’s five priority sectors, health occupies a unique position. It’s the first national mission and the first data space backed by real money and early opportunities to turn intentions into something tangible. That means $200 million for a first AI mission focused on improving health outcomes, $100 million for a national Health Sector Data Space with CIHI, and $100 million to expand the VITAL hospital data platform into five additional provinces.
$200M
for an AI mission on health outcomes
$100M
for a national Health Sector Data Space
$100M
to scale hospital data platforms
That concentration is not an accident, but rather the government quietly admitting that health is where the strategy’s central claim gets tested first and most visibly. The strategy’s own diagnosis is that Canada’s problem was never invention, but adoption. And if AI adoption is going to prove itself anywhere in Canada, it should be in healthcare.
Health is the proving ground. And the question worth asking before the funding announcements turn into procurement, is what it actually takes to make a proving ground prove something.
The Health Transformation Lesson Canada Gets for Free
I work for a company headquartered in the UK, and the UK has already run versions of this experiment with very different results. That history is a useful thing I can offer the Canadian conversation, because they were expensive lessons and they have already been paid for.
The UK’s record here isn’t unblemished. The NHS National Programme for IT launched in 2002 and set out to give every patient in England an integrated digital record. It was dismantled in 2011 and cost over £12 billion, a cautionary tale of what happens when ambition outruns delivery.
A decade later, another national health program ran into a different problem that parallels to the opportunity before Canadian health leaders today. That program was care.data. Launched in 2013, it set out to extract and link patient data from across the NHS into a central system for research and planning, which is precisely the category of initiative that VITAL and the CIHI Health Sector Data Space represent. The care.data program was put on hold in 2014 and eventually cancelled, and it did not fail on technology. It failed on trust. Patients were never clearly told what would be done with their data, the program was opaque about who would get access, and the prospect of data being sold to commercial third parties, including insurers, turned a reasonable idea into a public scandal. The lack of genuine informed consent was the central reason for failure.
The analysis reads like a design brief for what comes next. The lessons drawn from care.data were the need for clear communication to the public, consent rules in plain language, and strong oversight of who is allowed to use patient data and why. Researchers studying the collapse concluded that the support of clinicians should have been earned before implementation rather than after. Another key finding was that confidentiality, consent, and trust needed to be addressed in the planning phase rather than treated as things to manage once the system was live. None of those are technology problems. They are delivery and design problems, and they are precisely the failure modes that a well-funded, ambitious, nationally directed data strategy is most exposed to.
I raise it not to be discouraging, but because the contrast with what the UK did next is the whole point.
Building Trust in Digital Healthcare: Lessons from Genomics England
While care.data was collapsing, another national health data platform was being built that took the opposite approach, and it worked. That platform is Genomics England, and it succeeded for reasons that have everything to do with delivery and design, and almost nothing to do with the strategy on paper.
Genomics England is owned and funded by the Department of Health and Social Care, and it was built around four aims that are strikingly relevant to what Canada is now standing up:
- To create an ethical, transparent program grounded in participant consent.
- To deliver tangible benefit to patients through an NHS genomic medicine service.
- To enable new scientific discovery.
- To seed a domestic genomics industry.
Sovereignty, consent, patient outcomes, and economic development were all named as design goals rather than afterthoughts. Where care.data assumed public trust and lost it, Genomics England set out to earn it.
The architecture reflects that. Approved researchers don’t extract the data—they come to it. They work inside a secure research environment where information is de-identified and re-identification is prohibited. The data never leaves. This “bring the researcher to the data” model is the same principle that VITAL’s “living laboratory” framing and the CIHI data space both imply. It is precisely the safeguard the post-care.data literature now prescribes as the way to share sensitive health data without betraying the people it came from. It is also the hard part, because it must balance genuine research utility against genuine privacy—at population scale and while keeping public trust intact.
That trust was earned rather than assumed. The program was built on an explicit, public, consent-based framework with participants, and it has been challenged and scrutinized along the way. That scrutiny is a feature, because a national data platform that hasn’t had to defend its consent model in public hasn’t yet been tested.
There’s a delivery story underneath the mission story. And it’s where Kainos and Davis Pier do our best work. When Genomics England needed to move its national research platform, comprising tens of petabytes of genomic data, onto modern cloud infrastructure, the early going was hard in an instructive way. The initial platform was actually less performant than the on-premise system it replaced, and researchers were forced onto the largest, most expensive compute just to get adequate results. That is the detail every Canadian health leader should sit with, because it shows that funding a platform, and even building one, does not guarantee it works. Kainos was brought in to fix it, and the migration that followed moved roughly 50 petabytes to AWS, delivered about a year ahead of schedule, cut projected running costs by around 90%, and reduced key processing jobs from 25 hours to 23 seconds.
90%
reduction in running costs
23s
to run key processing jobs
I share those numbers not as a highlight reel, but because they make a single point. The distance between a funded national data platform and usable one that researchers can actually do science on is a delivery chasm, and strategy documents don’t cross it. Delivery does.
What Canada’s Health Sector Data Space Needs to Succeed
Canada has just funded its own versions of this, with VITAL scaling a proven Ontario hospital data network nationally, and the CIHI Health Sector Data Space linking standardized datasets for research and system improvement. Both are the right moves, but both are also where the UK was at the start of its journey rather than the end. The Canadian programs are governed differently from care.data, with no equivalent appetite for selling data to commercial interests, and that’s to their credit. But governance on paper and trust in practice are not the same thing, and the latter must be built deliberately.
The strategy treats adoption largely as a problem of literacy and access and frames it around teaching more people and connecting more data. In the health sector, that framing is incomplete because adoption is really a problem of workflow and trust. A tool that doesn’t fit how a clinician works won’t be used, however capable it is. A data platform that hasn’t earned the public’s confidence won’t keep its social licence, however well-funded it is. And a national data asset that can’t get its insights out of the research environment and into the point of care produces papers rather than better outcomes. The last mile is the whole game.
None of that is a criticism of the strategy. It’s a description of the work the strategy can’t do for itself, and the work that determines whether the proving ground proves anything.
The UK has shown both how this fails and how it succeeds, and the difference between the two was never the ambition or the funding. It was whether someone did the foundational work of designing for consent from the outset, redesigning workflows around clinician behaviour, earning trust in the open, and closing the last mile to the patient.
Canada doesn’t have to relearn those lessons the hard way, it can inherit them. That’s the opportunity in front of us, and it’s the work that starts now.
Ben Goldberg is a Principal, Health Industry Advisor at Kainos. He works closely with Davis Pier, leading healthcare market growth across Canada. Ben brings 25+ years of experience in IT delivery and business development across government, hospitals, public laboratories, community care, and private industry, with deep expertise in health system transformation.
