Palantir isn’t magic, but their staff are well paid to work hard to make Palantir irreplaceable. The Commons Science Innovation and Technology Committee recognises this is a risk, and it has been left to a new Health Secretary to figure out what to do next.
We offer a workable plan to mitigate the (multiple) risks, and to create options for the NHS which can increase accountability and trustworthiness.
Whether moving off Palantir is possible isn’t meaningfully contested. A “former FDP lead” agrees that Palantir can be replaced: “From a technology perspective, there is no lock-in that you couldn’t get out of.” There’s no dispute such a move could succeed – the CEO of Palantir UK agrees “there is no technical lock-in”, although not all forms of lock-in are “technical” – but there are different views on whether it should be done, and whether an attempt would be successful.
At this point, even trying to ensure a level playing field will generate further profit-motive derived subversion and novel PR campaigns – perhaps akin to an ex-PM commenting on current Labour leadership – but the NHS refusing to create options will stoke even more public opposition and continue undermining public trust. The NHS does not exist as a cash cow for suppliers; it exists to provide care – and citizens, patients, and voters all get a say in how that happens.
The NHS frontline ‘plays on hard mode’; if a patient says no to a blood transfusion, they can do that. Technology for care can’t ignore the constraints that the frontline operates on, even if it would make staff or suppliers’ lives easier.
Palantir’s UK lead says both that the Government’s digital ID programme “isn’t one for” Palantir and that he “would not want to live in a world where unelected tech bosses can decide which government policies can actually be delivered” – speaking out on politically relevant projects he wants to speak on, while remaining silent on issues where he could make identical claims that don’t need a large cheque upfront, or the clunking fist of state-mandated Surveillance Nationalism.
Speaking on Radio 4, Palantir’s UK lead said: “The proposal is that the NHS cancel our contract, but it does not propose an alternative” – because of course he’d say what’s in his commercial interests. Such a plan would come out of the Health Select Committee, which isn’t the one that wrote this report; medConfidential are glad that Mr Mosley seems to agree that such a report should be written. Any pathway has to be grounded in honesty to patients.
When discussing the Palantir platform, DH/E told the National Data Guardian that “access to identifiable patient information would be limited to NHS staff with a legitimate need” – then the NDG discovered Palantir’s staff had ‘unlimited’ access as well, by reading it in the press. That Palantir staff have access is one thing; that DH/E lied about it is quite something else.
Some Tech companies learn lessons, some don’t
The “P0 Plus Plus” chapter in the recent DeepMind book has a thought experiment on what might have happened with different leadership, and the entire book shows how it would have been different under Demis. What it shows is that it’s not always that someone can’t do what the company wants to do, but rather they can’t because they lack public acceptability due to decisions they made in the past. Those decisions may have been opportune, and got them to where they are – but now they prevent them going any further. Sometimes one company can’t do everything under a single division, with a single person leading everything.
Palantir suffers from the same Tech Bro delusion – their past choices have closed off public acceptance of some future options. They may think they can do anything, but that doesn’t mean they can do everything.
Conversely, something which under one regime is a weakness can become an opportunity under renewed political leadership, where good options can be implemented – the biggest barrier is haste over speed; the more you rush into things, the slower the progress.
All of Palantir’s funding that went to Peter Mandelson, and all of the pre-election ‘education’ of Wes Streeting count for much less than they did on the day before he quit; Streeting hasn’t left any even mediocre options on the table. Wes may say “We learned at terrible cost in Iraq what happens when loyalty replaces judgment”, but the same argument applies to his own ongoing loyalty to Palantir and the Peter Mandelson playbook.
The long goodbye will be sabotaged by political games; a quick quit is currently difficult because it was designed to put patients at risk. To move forward, that risk must be managed first, and removed, before the platform is (potentially) replaced.
Palantir can hold cancer patients hostage
Palantir claims their tech is better than anyone else’s – if so, they should be willing to prove it in the market and not rely on lock-in, monopoly, and bluster.
Cancer360 (the Palantir-only cancer pathway tool) and Optica (the Palantir-only hospital discharge tool) are the tools by which Palantir implements their vendor lock-in monopoly. The NHS owns no intellectual property on Cancer360; Palantir decides how and where it works. An intern at Palantir has more unilateral authority over every cancer pathway in the country than any NHS cancer doctor – this is the system as it exists today.
Turning off Cancer360 means breaking cancer pathways (as FDP advocates admit), and that’s not an option.
NHS England should immediately require that there are three competing equivalent tools for cancer hospitals to use to manage their pathways. This will provide resilience, it will help competition, and it will help Trusts work how they want to work – and crucially it will remove the ability for a single supplier to dominate a key National Health function.
As is argued, “The Solution Exchange needs a commercial framework and early adopter engagement now”… “None of this is happening at the pace it should”. The only organisations who can put products on FDP right now are NHS bodies and Palantir. A flourishing platform of choice and innovation, with suppliers competing to meet specific needs and provide the very best products doesn’t exist. It should – it is what was promised – but it doesn’t. On the FDP as it exists today, any supplier competing with Palantir must pay Palantir.
As it is for cancer, so it is with Optica for discharge – although the F1/F2s who keep A&E running are quite used to national decisions making their life harder, so would likely cope with Optica going away temporarily.
Step one: DHSC/NHSE can refuse to allow new monopoly Palantir products until there are multiple competing suppliers for that product.
Competition will take a short time – the best time to start this was in 2024, the second best time is now – but competition ensures there’ll be no aspect of FDP that is eternally bound to Palantir.
Re-tender in future
With competitive suppliers for all of the functions on an NHS FDP, the Platform can be re-tendered as and when the NHS chooses. (Noting that creates a long overlap for the kind of dark arts that the Dark Mister Mandelson practiced, prior to his disgrace.)
The incumbent FDP supplier will have people in every meeting, and will spend whatever it needs in order to keep the underlying platform contract – but whatever is best for patients today and in the long term should win. Vendor lock-in can never be best for patients in the long term.
Of course, patient and public opinion will also have to be taken into account.
Maybe Palantir will say they’ll allow Cancer360 to run on Apache Spark platforms from their competitors. That is entirely possible, technically – and it provides Palantir with the choice to show whether they care more about patient care or their own profits and lock-in.
It is entirely reasonable for a tech company to choose either way. Apple produces its Music app for both Windows and Android, but doesn’t offer its Garageband application; these are purely business decisions for business reasons.
Like Apple, Palantir is allowed to choose – but the NHS should not be hostage to Palantir’s choices.
Equally, NHS Trusts should be able to choose the suppliers of their tools. Is there lock-in to FDP apps or not? If not, then they can exercise choice. If there is, then that’s a choice too…
From Palantir’s perspective
Fujitsu was the tech company of the future once. Palantir and their supporters’ understanding of their tech isn’t wrong – but the choices they make on how to use it are limited by Palantir’s culture, its ideological leadership, and their decisions. If selling out the public Horizon-style met Palantir’s interests, they would act just like Fujitsu.
And that is the underlying generator of lack of trust: that Palantir will do what is in US Palantir HQ’s interests, not the British public’s. Things were much simpler when Palantir’s customers were solely the US Military, who don’t care what those who are on the impact end of a bomb think about the bomb being dropped. The British public demand far more say than that over NHS decisions.
Palantir got this wrong at the start of the contract, and NHS England’s inaction meant Palantir never had to learn the lesson. Palantir exacerbate the worst instincts of the Department of Health in England: to centralise control and remove autonomy from professionals who understand nuance and local expertise.
Palantir only considers solutions that are entirely compatible with existing and expanding Surveillance Nationalism. When all you have is a hammer, every source of revenue must be treated as a nail – which is the sort of approach Peter Thiel has just fled to Argentina to escape
NHS Code should be portable (and can be made portable where corners were cut)
It’s a common fallacy that the only silo that matters is the one the advocate sits in. There are legitimate analytical needs across the full Reproducible Analytical Pipeline spectrum that should be facilitated. (RAP used to be an unpolluted acronym…) When IT Crowd Tom argues in favour of FDP, or addresses some criticism, that is not the same thing as arguing for Palantir.
He claims all (his) NHS written code is portable without lock-in, so his important but long-neglected basement team has nothing to lose from changing platforms. This is what he argues when he’s not making the conflicting claim that only Palantir can do FDP – Tom can’t have it both ways, no matter how much it makes his life easier to try. FDP is an expensive way to run commodity R/python code owned by the NHS, but it’s the only way to run proprietary R/python code owned by Palantir.
When the neglected deprived kids in a poor neighbourhood are tempted by a free trial and cheap opening offers from the local dealer, the right answer is not to have someone a bit less malign make the same offer. Instead, it’s to help them now – and those services that help them will help all those who follow.
Palantir isn’t magic. It’s just software, and good software is portable, but political decisions are political decisions.
As it operates today, the Platform can tell patients how data about them is accessed and used – Palantir’s FDP is entirely capable of doing that – but NHS England point blank refuses to provide that feature to patients, choosing instead to side with creeps and ghouls who look up the records of people they know and victims of terrorism. FDP could show each patient exactly where their data is used and for what – that logged audit trail exists in Palantir’s software, why isn’t it turned on?
Software can be written to do what is wanted, currently there’s no agreement on what that is.
Available Next Steps
Palantir isn’t a revolutionary force; it is an expensive cage built in software. The true threat to the NHS is not a lack of technology, but the willingness to prioritize vendors over patient dignity and choice.
The way forward is not an abrupt exit, but a firm insistence on competition that forces accountability. Because when the health of a nation can be managed by the deepest pockets and the slickest playbooks, the patient (and the electorate) hold the final veto.
Step 0: Turn Palantir’s audit trail into patient rights. Immediately enable NHS patients to see exactly where their data is used now – via FDP’s existing logs, to all patients who have automatic access to GP documents in the NHS App – not as a tech feature, but as a public trust mechanism. This should also include GPConnect and Summary Care Record audit logs held by NHS England as an immediate step to track ghouls and creeps.
Step 1: Immediately require 3 competing cancer pathway tools (not just one Palantir monopoly) for NHS hospitals – breaking Cancer360’s lock-in next to prevent using cancer patients as hostages.
Step 2: Mandate competition for all FDP products – no new Palantir tools can launch outside pilot hospitals until at least 2 other suppliers (including non-Palantir) prove they can handle the same workflows.
Step 3: Force re-tendering with competition – NHSE must re-run FDP contracts only after competitive suppliers prove they can deliver better (not just cheaper) patient outcomes than Palantir. Suppliers can provide the Platform or Products, not both.
==
Find out what happens next: Sign up for our newsletter (we don’t email often) or get small frequent updates via our free Substack – updates are free to read; we’re very grateful to all who can support our work with a Monthly or Annual subscription, or make a regular or one-off donation.