The Single English County Saying No to Palantir
Greater Manchester, a major NHS metropolitan region covering around 3 million people, has refused to adopt a national federated data platform (FDP) built by Palantir, preferring its locally developed Analytics and Data Science Platform (ADSP). That refusal has become a focal point in a national debate about vendor concentration, public trust in health data, and how to scale AI and analytics across a public health system.
According to WIRED (Aug 21, 2026), the UK government has roughly six months to decide whether to activate a break clause in a Palantir contract “worth more than $400 million.” If the contract is not terminated, the deal could run “until 2031.” WIRED also reported that Palantir was commissioned in 2023 and that a national rollout began in early 2024. Separately, NHS England’s FDP uptake pages (end‑May‑2026 snapshot and product-level reporting to end‑March‑2026) publish sign-up and product metrics with an explicit methodological caveat: their observational before/after figures cannot establish causation.
FDP vs ADSP, what these platforms actually do
Federated data platforms let identifiable records remain under local control while enabling pooled analytics and shared tools across regions. Palantir’s FDP is designed as a national pool plus local instances so tools developed in one area can be “lifted and shifted” to another.
Greater Manchester’s ADSP is a locally built, modular collection of tools the region says it has developed over roughly a decade. The Integrated Care Board (ICB) has argued the ADSP is configurable, includes primary‑care data that the FDP allegedly lacks, and, at a May 2025 board meeting, concluded its “local capability exceeds anything the FDP currently offers, ” with some functionalities “two, three years” ahead, according to reporting.
What adoption looks like, and why the numbers need careful reading
Numbers are central to the argument, but they measure different things.
- WIRED reported that “139 of roughly 200 trusts currently ‘live’ with Palantir’s technology, ” and that “35 of England’s 36 ICBs are actively using the system.”
- NHS England’s FDP uptake page (end‑May‑2026) reports that 35 ICBs had signed a Memorandum of Understanding (MOU) to join the FDP. That is a useful, dated data point, but an MOU is not the same as daily operational telemetry.
- NHS England’s product reports (to end‑March‑2026) show per‑product counts such as 37 trusts using the Inpatient Care Coordination Solution waiting‑list module and 39 trusts using its theatres module, plus cumulative impact figures (e.g., 111, 589 additional patients undergoing procedures in theatres; 87, 842 patients safely requested for removal from inpatient waitlists; 218, 389 patients safely requested for removal from outpatient waitlists).
NHS England: these before/after benefit calculations cannot establish cause-and-effect.
Two points flow from this. First, “signed up” (MOU) is a contractual or engagement metric. “Live” can mean anything from a single pilot module to full daily usage. Second, NHS England’s operational impact numbers are observational. They show activity and correlation, not definitive proof the FDP caused those outcomes.
The core arguments
Why Greater Manchester rejected the FDP
Trust and clinical legitimacy are central to Greater Manchester’s position. Matt Hennessey, chief data and analytics officer at NHS Greater Manchester, framed the concern plainly:
“[Even] a technically strong platform will struggle to realize value if clinicians, data controllers, patients or the public do not trust it, ” said Matt Hennessey. “If we were to fully adopt the FDP … it would be a retrograde step.”
Greater Manchester also emphasises modularity. Hennessey argues that because the ADSP is “a collection of technologies, ” the region can swap out components if they fall behind. He also claims the ADSP includes primary‑care datasets not available in the FDP, a practical difference for many local commissioning and patient‑pathway decisions.
Why proponents want the FDP to continue
Proponents stress scale, portability, and the promise of a single surface for AI tooling. Tom Bartlett, an independent IT consultant and former deputy director of data engineering at NHS England, summarised those benefits:
“You can lift and shift. That’s the real power of the FDP, ” Bartlett said. “The other advantage is that you’ve got a surface for artificial intelligence to work across.”
Palantir UK’s Stephen Childs has pointed to frontline users: “Thousands of doctors, nurses, and other NHS staff use the [FDP], with many on the record as to its benefits, ” he told reporters. For those backing the national route, rolling back a single‑vendor programme risks losing years of shared tooling, interoperability work, and a platform intended to accelerate AI and analytics at scale.
Politics, reputation and public trust
The debate is as much political and reputational as it is technical. Palantir’s work outside health, including defence and immigration projects, and public statements by cofounders have fuelled opposition. Reporting highlights several flashpoints: Palantir cofounder Peter Thiel suggesting the UK should “rip the whole [NHS] from the ground and start over, ” a British MP calling a Palantir manifesto “the ramblings of a supervillain, ” and academic critics questioning whether Palantir’s corporate stance fits public‑sector data values. As Jessica Morley, a health data researcher at Yale University’s Digital Ethics Center, put it: “Palantir is essentially antithetical to all of those values.”
That reputational layer has prompted parliamentary scrutiny. WIRED reports a bipartisan group of MPs published a report in June 2026 warning national reliance on Palantir is an “unacceptable point of weakness, ” and that a parliamentary committee in July made similar arguments drawing on testimony from Matt Hennessey. WIRED also reported the government has a contractual break clause window that could be exercised within the timeframe described above.
Where the evidence is thin, and what leaders should demand
The public arguments raise real operational and governance questions that don’t yet have public, independent answers. Executives, CIOs, and board members should press for them before committing to any national platform.
- Independent, peer‑reviewed comparison. There is no publicly available, comprehensive, peer‑reviewed study demonstrating ADSP’s superiority in clinical outcomes or value‑for‑money at national scale. Greater Manchester asserts functional advantages; Palantir and NHS England point to uptake and activity figures. That gap matters.
- Clear definitions of “signed up” vs “live.” Ask for precise telemetry: how many clinicians log in weekly; which modules are used and to what depth; and whether product usage is pilot‑level or operationally embedded.
- Contract transparency. WIRED reported a contract figure “worth more than $400 million” and a break‑clause timeline; procurement documents should confirm contract value, start dates, the exact mechanics of the break clause, data‑residency clauses, and whether any national security exemptions apply.
- Contingency and transition planning. If a break clause is triggered, who maintains patient‑critical services, who owns the data export process, and how long would any migration take before service degradation risks emerge? Bartlett warned, “If the break clause got triggered, we’d go backwards, because there is no alternative, ” and added, “The trusts that were previously using paper would go back to paper.” This is a cautionary scenario that requires tangible transition plans.
- Separation of ethics vs engineering risks. Reputational concerns (defence/immigration contracts, executive statements) and technical concerns (data scope, modularity, audit logs) require different mitigations. Treat them separately but address both in procurement and governance checks.
A practical checklist for leaders negotiating large AI/data platform contracts
- Demand the contract or a machine‑readable summary that includes: data residency and ownership rules, deletion timelines, emergency exit triggers and timelines, fees and obligations for data extraction, and audit access for independent reviewers.
- Require independent, reproducible evaluations before scale roll‑out: third‑party clinical outcome audits, economic assessments, and security reviews.
- Insist on clinician-facing audit trails and explainable‑by‑design features for any AI/decision support tools; public trust requires both transparency and educational plans for staff and patients.
- Secure operational telemetry during any pilot: weekly active user counts, use-case depth (which modules and workflows), and measurable KPIs tied to clinical outcomes, not just dashboard views or clicks.
- Obtain detailed contingency plans: interim service SLAs, data extraction processes, and a timeline for transition, with named accountable teams and funded exit tasks.
Key takeaways, quick questions and honest answers
- Is Greater Manchester the only English ICB refusing the FDP?
As reported by WIRED (Aug 21, 2026), Greater Manchester is the sole English ICB that has repeatedly refused to adopt the FDP and continues to use its ADSP. That status is a public, dated claim and should be checked against local ICB minutes for any changes. - How many ICBs have engaged with the FDP?
NHS England reports that 35 of England’s 36 ICBs had signed a Memorandum of Understanding to join the FDP as at end‑May‑2026. An MOU indicates formal engagement but does not necessarily mean full operational use. - Do NHS England’s uptake figures prove the FDP caused better outcomes?
No. NHS England’s product-level counts and cumulative impact figures (for example, 111, 589 additional theatre procedures; 87, 842 inpatient removals; 218, 389 outpatient removals, cumulative to end‑March‑2026) are observational. NHS England explicitly warns these before/after calculations cannot establish cause‑and‑effect. - Is the Palantir contract at risk?
WIRED reported the government faces a window to trigger a contractual break clause (roughly six months from the report). Contract details and exact dates should be confirmed in procurement documents on government portals or via FOI requests. - What should health IT leaders focus on now?
Ensure transparent contract terms (especially data access, residency, audit and exit clauses), commission independent technical and clinical evaluations, require clinician‑level audit trails for any AI surface, and secure robust contingency plans to protect continuity of care.
My read, what this means for business leaders
This dispute is a governance stress test for national AI and analytics policy. The choices are not purely technical: they are about who holds the social licence to use patient data, how vendor risk is managed at national scale, and whether centralisation or local modularity better supports trustworthy AI agents in health care.
If you run a trust, a regional commissioner, or are a supplier positioning for national scale, treat Greater Manchester’s stance as a governance playbook. Do three things before you sign or scale:
- Get the contract and demand machine‑readable clauses for data access, emergency exits, and audit rights.
- Require independent, reproducible evaluations of clinical and operational impact before full roll‑out.
- Prioritise clinician trust: invest in explainability, audit trails, opt‑outs, and public communication that addresses reputational concerns head‑on.
At stake is what platform will host AI agents across the NHS. National scale brings benefits: shared models, cross‑region learning, and a surface for AI innovation. But scale without trust, transparency, and contingency planning risks public backlash and operational fragility. Greater Manchester’s “no” is not merely parochial stubbornness, it is a reminder that technology projects that touch patient data must win clinicians’ and the public’s confidence as well as technical benchmarks.
For executives, the practical lesson is simple: do not outsource governance. If you plan to adopt national AI platforms, insist on contractual clarity, independent evidence, and clinician‑centred safeguards, or preserve local alternatives until those conditions are met.