An analysis that never states its opponent's best case is advocacy. Here is the strongest version of the argument for "once for Wales" — stronger than its proponents usually make it — and what survives contact with the evidence.
The steelman
- Scale. Wales has 3.16 million people. Seven boards procuring separately means seven procurement overheads, seven integration projects, seven times the vendor margin. One national body amortises everything.
- Scarce talent. Wales cannot staff one great digital organisation, let alone eight. Concentrate the talent or dilute it to nothing.
- Single accountability. When one body owns delivery, ministers know whom to hold to account. Federate it and every failure becomes everyone else’s fault.
- Consistency. One patient, one record, one standard — hard enough with one builder; impossible with eight.
- Market reality. Health-IT vendors barely price a Welsh national deal; seven sub-scale deals get worse terms or no bids.
These are serious arguments. Two of them are partly right.
What the blueprint concedes
Some functions are genuinely national. Identity, the patient index, terminology, the interoperability backbone, cyber defence — duplicating these seven times would be waste. That is exactly why the target architecture keeps them national, in the standards body. The blueprint is not decentralisation; it is the correct boundary between what must be done once and what must not.
Talent is scarce. Which is the argument against the monopoly, not for it: scarce talent goes where it can ship. The record of the concentrated model is an 80% headcount increase and technical leaders leaving; the people plan is how a federation competes for talent that a discredited monopoly cannot attract at any scale.
What does not survive
Scale. The efficiency argument assumes the single body is efficient. The observed unit economics of the Welsh monopoly are on the record; Denmark and Estonia achieve the promised economies through shared standards, at a fraction of the cost, while their delivery layers compete. Scale economics live in the standards, not the delivery monopoly — the comparators are the controlled experiment.
Single accountability. In theory, one throat to choke. In practice, the funder co-authors the failure and the accountability framework itself was captured — a monopoly is harder to hold accountable, because there is no comparator and no exit. Seven boards on the same standards produce seven comparable delivery records; accountability gets easier, not harder.
Consistency. One patient, one record — delivered today in Denmark and Estonia by federation, and not delivered in twenty years by the Welsh single-builder model. Consistency lives in the data standard; forcing it through a single application builder is how NPfIT spent £10 billion learning the same lesson at English scale.
Market reality. Partly a real constraint — which the blueprint handles as designed: joint procurement as a board choice, coordination with NHS England where the market is genuinely UK-shaped, and international precedent (five Danish regions of comparable size procure successfully within MedCom standards).
The honest residual
Federation has a real cost: seven boards must become capable buyers, and today they are not. The blueprint prices that cost — the capability standard, the transition plan — rather than denying it. The choice is not between a tidy centre and a messy federation. It is between a model whose costs are visible and priced, and a model whose costs are £600M in, £0.5M out and structurally unpriceable, because nothing exists to compare it against. Priced beats unpriceable.