NHS Data Silos Aren't a Technology Failure. They're a Business Model.

When those plans fail, it's easy to blame bad luck, outdated software, or the fact that decades of local buying decisions left every hospital running something different. The natural assumption is that modernising the technology will make the problem go away.

That explanation is incomplete. It treats the fragmented IT systems like an accident of history. They aren't. They're exactly what you'd expect from the market that built them, and until that's the starting point, every fix keeps solving the wrong problem.

What NHS data silos actually are

Hospitals don't build their own systems. They buy them from a highly competitive market of private suppliers. And when they buy, they get locked in.

Take Stockport and Tameside and Glossop, which both signed new EPR contracts in March 2026. Like most modern NHS IT deals, they're locked in for a decade. Once the system is deployed, staff are trained and clinical workflows are built around it, the switching costs are enormous. Re-procurement rarely happens.

Think about it from the supplier's perspective. They have a ten-year contract, and their main goal is making sure the trust renews. If they build an easy bridge to a competitor's system, they are handing that competitor the keys to win the next tender. Nobody in a boardroom has to decide to keep a rival locked out. Building open, sharing systems simply works against their own commercial interest.

A GP with two decades of NHS experience put it to me plainly, unprompted:

Different private enterprises have got in and won tenders or contracts, so then we're kind of locked into their system, but then there's another system and then they won't talk to each other. You might have an incentive to not make their system lock up too much, 'cause then you call the market, your system's invaluable. We get a new system in, it's kind of the wrong system, but then we're stuck with it 10, 15 years.

For a clinician dealing with this daily, it feels intentional. It's worth being precise about why it isn't. The NHS didn't design it this way on purpose, and suppliers aren't being malicious. It's closer to physics. Given the financial incentives at play, this is the exact shape the market was always going to take.

Why the usual fixes keep failing

Two fixes get tried, and both have already failed at scale.

The first is to rip it all out and replace it with one government-built system. In 2002 the government launched the National Programme for IT and went on to spend £2.7 billion trying to deliver a single electronic care record across the NHS. The National Audit Office's 2011 review found it did not represent value for money, with no grounds for confidence the remaining £4.3 billion budget would fare any better. The government scrapped the programme that September, handing local IT decisions back to individual trusts. Replacing the market with a central authority isn't a hypothetical failure. It's recorded history, at a scale that makes it hard to attempt again.

The second is the opposite instinct: instead of replacing everything, let one supplier become the layer everything else has to sit on top of. This is happening right now. In June 2026, a cross-party Commons select committee warned the government to be ready to use a break clause rather than deepen the dependency on a single dominant supplier. This approach doesn't solve the problem: it swaps fragmented dependency for a different, and arguably worse, kind of risk.

Both fixes fail because they ignore the root cause. Ripping and replacing assumes the technology is the problem. Relying on one dominant supplier assumes the problem is having too many suppliers. Neither recognises that NHS data silos are a rational business response to how the contracts are written.

The third way: change who the data works for

There is a position between those two. It doesn't unify the market, replace the suppliers, or rewrite the procurement rules. It is a layer that makes the existing market work for the NHS rather than for the suppliers themselves.

Here is what that looks like. The current suppliers keep their contracts, their hardware revenue, and their hospital relationships. Nothing gets ripped out: the systems hospitals have just spent years deploying stay where they are.

A governance layer sits over the top, inside the hospital's own walls, and makes the data the trust already holds readable and usable for its clinicians. The data starts working for the trust and the clinicians who generated it, instead of being legible only inside whichever supplier's system won the last tender.

To be clear, this isn't another monopoly chokepoint in disguise. The test is simple. Does it keep competition between suppliers alive and only change who the data serves? Or does it become the one thing every supplier and every trust has to route through? The first is a genuinely different fix. The second is the single-dominant-supplier problem again, in different clothes.

It's an incentive problem, not a tech problem

None of this needs a new file format or a better standard. NHS systems have failed to talk to each other through several generations of technology because the incentives never changed. What has to change is who the data works for at the moment it's captured, one hospital building at a time, without dismantling the market to do it. That doesn't depend on a new mandate or a regulator forcing the issue. It needs the incentive on the ground to point the right way.

It matters now because trusts are adopting AI at pace, on top of exactly these systems. Whatever an AI agent is given, it inherits the data as it stands, silos included. Put one on top of a fragmented estate and it doesn't fix the fragmentation. It automates it.