Palantir isn’t just selling software, it’s quietly capturing the state

As Palantir embeds into government operations, the real risk is not efficiency, but dependency and weakened democratic oversight.

Article link: https://open.substack.com/pub/johanosteyn/p/palantir-isnt-just-selling-software?r=73gqa&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

Audio summary: https://youtu.be/lplt6r1aNF4

Follow me on LinkedIn: https://www.linkedin.com/in/johanosteyn/

There is a line between a government buying technology and a government becoming dependent on a private platform to function. My worry is that the UK is drifting across that line with Palantir. The story is often framed as modernisation: better data, better decisions, faster services. But when the same vendor sits inside defence, health, and other high-trust domains, the issue stops being procurement and starts becoming power.

Once workflows, data pipelines, and institutional habits are built around one system, exit becomes hard, scrutiny becomes thin, and democratic accountability starts to feel like an afterthought. And if the UK, with all its capability, struggles to hold this relationship at arm’s length, it is a warning to every country tempted to “buy a state operating system” rather than build state capacity.

CONTEXT AND BACKGROUND

In early February 2026, The Guardian reported growing calls to halt or pause UK public contracts with Palantir amid concerns about transparency and the expanding footprint of the firm across state functions.

The defence side of the story has become harder to ignore. The Register described a direct-award contract worth roughly £240 million for the Ministry of Defence to continue licensing and supporting Palantir’s data analytics work, framing it as a major “follow-on” agreement.

Around the same period, openDemocracy examined what it described as a revolving-door dynamic between the MoD and Palantir, arguing that the relationship is deepening at the same time as the company’s role becomes more strategic.

INSIGHT AND ANALYSIS

Palantir’s defenders will say the obvious thing: governments are overwhelmed, legacy systems are fragmented, and better tools help public servants do their jobs. That is not an absurd argument. The problem is what happens after the tool becomes the backbone. Once a platform is embedded, it starts to shape how the institution sees the world. It influences what gets measured, flagged, prioritised, and acted upon. Over time, “decision support” quietly becomes decision-making, simply because people trust what the system surfaces and ignore what it cannot.

The second problem is lock-in. It is not only commercial lock-in, where switching becomes too expensive. It is skills lock-in (the state loses the capability to build and manage alternatives), process lock-in (workflows are redesigned around the vendor), and operational lock-in (critical functions can’t run without the platform). At that point, procurement becomes a fiction. The state is no longer choosing; it is renewing.

Health data adds another layer: trust. The NHS is not just a service; it is a national trust relationship with citizens. When the public believes their data is being routed through a controversial foreign vendor, the long-term cost is not only privacy risk, but legitimacy risk. In 2026, the BMJ reported that doctors would be given guidance on limiting engagement with the NHS Federated Data Platform because of links with Palantir, a signal that professional unease has not evaporated.

The final concern is geopolitical. Even if contracts promise “sovereign” control, dependence on a US-based firm inevitably creates a legal and strategic exposure that the public rarely understands. You do not need a conspiracy theory to see the leverage: states use law, diplomacy, and intelligence relationships to pursue national interests. The more mission-critical the system, the more valuable the leverage becomes.

IMPLICATIONS

For the UK, the first implication is simple: if Palantir is now woven into defence planning and core public infrastructure, oversight must be proportionate to the risk. That means transparency where possible, clear lines of accountability, independent audits, and a genuine ability to exit. The fact that the relationship is being framed as “strategic partnership” should raise the bar, not lower it.

For citizens, the implication is about consent and trust. When a platform becomes central to the state’s operations, people deserve more than reassurance. They deserve plain-language explanations of what the system does, what it does not do, who controls access, and what safeguards exist against mission creep.

For countries like South Africa, the lesson is urgent. If a highly capable state struggles to prevent a vendor from becoming infrastructure, a developing state with constrained budgets and skills will be even more vulnerable. Digital modernisation is necessary, but outsourcing sovereignty is not. The long game must include building in-house capability, insisting on open standards, and designing systems so that the state remains the operator, not the tenant.

CLOSING TAKEAWAY

Palantir may well deliver real operational benefits, and it would be naïve to pretend governments can simply “opt out” of data-driven systems. But my worry is about the direction of travel: when one company becomes too embedded across the machinery of the state, the balance of power shifts quietly, and reversing that shift becomes politically and technically painful. The UK should treat this as a sovereignty question, not a procurement question. And the rest of us should watch closely. In the AI era, the most important infrastructure may be the systems that tell governments what is happening, what to fear, and what to do next.

Author Bio: Johan Steyn is a prominent AI thought leader, speaker, and author with a deep understanding of artificial intelligence’s impact on business and society. He is passionate about ethical AI development and its role in shaping a better future. Find out more about Johan’s work at https://www.aiforbusiness.net