What Happened
A detailed analysis published by Health Services Daily on 19 June 2026 argues that Australia's plan for a world-class national health data network is in genuine danger of missing the AI window entirely. Written by Jeremy Knibbs, the piece does not dispute Australia's structural advantages. It acknowledges the legislation, the system architecture, and a committed group of public servants working through a well-laid-out plan. The raw material is, in Knibbs's framing, arguably the best of any country on the planet: a unified national health identifier and 24 million citizens already enrolled in a pre-linked digital health record through My Health Record.
The problem, as the article frames it, is speed relative to what is already happening elsewhere. What Australia is thinking might take at least three to five years to get up and running, the Mayo group in the United States has up and running already in that country and many others. That gap is not a minor scheduling issue - it is a structural risk to the entire investment.
Why It Matters
The stakes here are clinical, not just administrative. Breakthroughs in diagnostics, preventative algorithms and treatments, genomics and even new drugs that previously took over 15 years are now being achieved in weeks and months. Every year of delay in building a functioning national health data network is a year in which Australian patients miss access to those tools.
The article's title - "slow has a body count" - is blunt for a reason. Governance caution is not a cost-free default. The real risk calculation, as Knibbs puts it, is not "could this AI system ever give wrong advice?" but rather "is what we try safer than the realistic rapidly evolving alternative?" That reframing matters enormously for how Australian health agencies should be assessing procurement and compliance decisions right now.
For organisations thinking through their own AI strategy, the health sector example is instructive: the cost of inaction compounds in ways that a standard risk register does not capture.
Key Details
The financial picture in the article is striking. Australia spends approximately A$1.9 billion each year on Primary Health Networks - some high performing, some wasting large chunks of money on ill-conceived, sometimes duplicated pilots. St Vincent's was recently awarded a government tender worth A$400 million to develop a low acuity national mental health app and program. Against those figures, the Mayo group's cost estimate looks very different: for an individual hospital with a couple of million patients, Dr Halamka estimates engagement costs between US$1-2 million, with a nine-month lead time from initiation to functioning analytics.
The institutional response to that comparison is predictable. If the Mayo platform's proposition - that a motivated Australian organisation could be up and running in nine months for US$1-2 million - was brought to an agency currently mid-way through a billion-dollar replatforming contract, the response would be: "The contract is signed - the ship is already under sail - pivoting now would be foolhardy."
That logic is understandable at the individual project level. Across a national system, it is how windows close.
The article also challenges how risk is framed in health AI governance. Rather than asking whether an AI system could ever give wrong advice, Knibbs proposes a more useful question: "what are the components of safe driving, and how do we set standards around each of them independently?" That approach - decomposing risk rather than treating AI as a single undifferentiated hazard - is closer to how regulators in other sectors handle complex technology. Organisations working through AI governance and training will recognise the same tension between component-level safety and system-level paralysis.
Background and Context
Australia's digital health infrastructure has been built over many years. The My Health Record system, administered by the Australian Digital Health Agency, gives Australia a pre-linked record base that most countries cannot replicate quickly. The OAIC oversees privacy obligations under the Privacy Act 1988, and the Australian Consumer Law framework (see en.wikipedia.org/wiki/Australian_Consumer_Law) shapes how data-driven services must treat consumers.
The challenge Knibbs identifies is not that these frameworks are wrong. It is that they were designed for a slower-moving environment. Government doers are caught trying to run a truly transformational idea within the constraints of old healthcare system dynamics - in particular risk, compliance and procurement dynamics. The article notes that one official quoted in the piece acknowledged the situation with a line that captures the cultural problem precisely: "Yeah, Kevvy's a great bloke - shit coach though." The implication is that individual capability is not the constraint; the system around those individuals is.
The priority now, as one source in the article puts it, is "building on these foundations to drive innovation in how data is used and continue delivering meaningful, nationally consistent insights." The question is whether the current pace of institutional change is fast enough to make that happen before the AI advantage shifts permanently to systems already operational overseas.
For Australian businesses thinking about AI automations in regulated sectors, the health data story is a useful reference point: the regulatory environment is real, but it is not static, and organisations that wait for perfect clarity tend to find the window has moved.
What Comes Next
The article does not call for abandoning governance - it calls for updating the risk question. The shift from "could this go wrong?" to "is doing nothing safer than trying?" is a meaningful one for any agency or health organisation currently in a planning or procurement cycle.
Practically, that means Australian health bodies should be looking hard at what the Mayo group has already built, stress-testing their own timelines against a nine-month/US$1-2 million benchmark, and asking whether their current spend on PHNs and large platform contracts is actually buying the capability the system needs. It also means the AI strategy conversation in health needs to move from "should we?" to "how fast, and in what sequence?"
Mindiam works with Australian organisations on exactly this kind of sequencing - from initial AI strategy through to implementation and automations. The health sector's structural challenge is an accelerated version of what most large organisations face: good intentions, real assets, and governance machinery that was not built for this speed. Read more about our approach and editorial standards.
Frequently Asked Questions
What makes Australia's health data position unusual compared to other countries?
Australia has a unified national health identifier and 24 million citizens already enrolled in a pre-linked digital health record through My Health Record. That combination - a single identifier plus a pre-linked record at national scale - is something most countries cannot replicate quickly, which is why the Health Services Daily analysis describes it as arguably the best raw material of any country on the planet. The infrastructure investment has been substantial and the legislative framework is largely in place or coming soon.
Why is the Mayo group comparison significant for Australian health agencies?
The Mayo group's platform is already operational in the United States and multiple other countries. Dr Halamka's estimate - US$1-2 million and nine months for a hospital with a couple of million patients to reach functioning analytics - sits in stark contrast to the three-to-five-year timelines being discussed for Australia's national build. That gap matters because AI-driven breakthroughs in diagnostics, genomics, and drug development are happening now, not in three years. Every year of delay is a year in which Australian clinicians and patients operate without tools that are already available elsewhere.
How should Australian health organisations think about AI risk differently?
The article argues that the standard risk question - "could this AI system ever give wrong advice?" - is the wrong frame. A more useful approach asks what the components of safe operation look like independently, and whether attempting a new approach is safer than the realistic alternative of doing nothing while the rest of the world moves ahead. That reframing does not remove the need for governance; it makes governance more precise and less likely to produce paralysis. Organisations working through AI training and governance design will find this distinction practically useful.
What does the A$1.9 billion PHN spend tell us about health system priorities?
Australia spends approximately A$1.9 billion per year on Primary Health Networks, with performance varying significantly across the network. The article uses this figure alongside the A$400 million St Vincent's mental health app tender to illustrate that the constraint on building a national health data capability is not primarily financial - it is about how existing commitments and institutional inertia shape what feels possible. The money is in the system; the question is whether it is being directed toward the highest-leverage investments.