The UK Just Fired Palantir. The Real Story Isn't the Contract.

The UK Home Office did something quietly radical this month: it ripped out the Palantir-built system it had been using to process asylum claims and replaced it with software written in-house. No press conference, no big announcement — just a swap. But for a government that has spent the better part of a decade handing critical infrastructure to a handful of US vendors, “we’ll build it ourselves” is a sentence with real weight.

How big Palantir had gotten in Britain

Palantir is not a peripheral player in the UK. It holds a £330 million contract to run the NHS Federated Data Platform, has long-standing ties with British defense and intelligence, and — until now — was the engine behind Home Office refugee processing. One company, three of the most sensitive data domains a government touches.

That concentration created an awkward asymmetry. When a single vendor sits between you and your own data, you end up asking the vendor how your data works. Ministers don’t love that, and neither do auditors.

Why the timing isn’t a coincidence

“Digital sovereignty” stopped being a think-tank phrase and became a political talking point across Europe over the past two years. The conviction is simple: routing core public infrastructure through US hyperscalers and US-aligned analytics firms is a geopolitical exposure, not just a procurement question.

Palantir specifically has been a lightning rod. Peter Thiel’s politics, the company’s deep ties to US agencies, and the opacity of how it processes data have drawn sustained fire from civil society groups and Labour backbenchers. Asylum data — by definition, information about some of the most vulnerable people in the country — was always going to be the hardest contract to defend.

“We built it” matters more than “we switched vendors”

The interesting word in this story isn’t replaced. It’s built. The Home Office didn’t migrate to AWS-flavored Palantir-lite from another vendor. It shipped its own system, leaning on the Government Digital Service capability that the UK has been quietly investing in since 2011.

Two reasons that matters. First, in-house builds compound — once you have the team and the muscle, the next system is easier to bring inside too. Second, this is now a precedent. Germany and France have been talking about sovereign govtech for years. A live, working example from a G7 government is the kind of evidence that turns talk into RFPs.

The bill comes due in year five

Building isn’t a free lunch. The hard part of public software isn’t the launch — it’s still running well in 2031. Patch cadence, on-call coverage, retaining senior engineers on civil service salaries against FAANG offers: this is exactly the territory Big Tech is good at and governments historically are not.

The Palantir trade was always partly a labor trade. You pay the markup, you skip the hiring problem. Walking away means the Home Office has to solve the hiring problem for real. Whether they can is the actual test, and we won’t know the answer for years.

The takeaway

This isn’t a goodbye to Palantir so much as a government deciding it would rather own the headache than rent the convenience. That’s a more expensive choice up front and a slower one. It’s also the only choice that ever changes your negotiating position.

Worth asking the same question about wherever you work: which conveniences have you traded away that would be genuinely hard to take back?