NHS Keeps Palantir, for Now

The UK government is preparing to extend Palantir’s £330 million NHS contract rather than trigger a break clause, according to the Financial Times, in a decision that says more about operational dependence than political comfort.

The contract covers the data platform meant to knit together fragmented health records, help hospitals discharge patients faster and improve surgery scheduling. Those are mundane tasks by the standards of AI marketing, which is exactly why this matters: if the plumbing sits deep enough inside a health system, swapping the supplier becomes a risk in its own right.

According to the report cited by The Guardian, a review concluded that forcing Palantir out of NHS data systems now would be too disruptive. Ministers instead plan a procurement process for a British alternative, with a possible replacement in two years.

That is a narrower claim than either side would like. It does not prove Palantir is the best long-term answer for the NHS, and it does not prove critics were wrong about handing sensitive public infrastructure to a US company with a highly political reputation. It shows something simpler and more uncomfortable: once a government buys a system to connect messy legacy data, ideology tends to meet the invoice.

Health Secretary Yvette Cooper has faced pressure from MPs and campaigners to end the arrangement. The objections are not just technical. Palantir’s work with the Israeli military and US immigration enforcement, along with founder Peter Thiel’s politics, have made the company an unusually combustible contractor for a public health service.

Palantir’s UK and Europe chief executive, Louis Mosley, responded in terms one would not normally associate with hospital booking systems, saying ending the contract would mean “putting politics above patients” and accusing critics of “Palantir Derangement Syndrome”. When a supplier reaches for culture-war language, it usually means the technology alone is no longer doing the persuasion.

The government’s apparent compromise is to keep the system running while trying to reduce strategic dependence later. Whether that produces a genuine British alternative or just a slower version of the same dependence is another question. Procurement processes are good at generating paper, less so at producing working software at national scale.

For now, the confirmed development is less glamorous than the argument around it: ministers appear unwilling to test how replaceable Palantir really is inside the NHS.