The NHS says its Federated Data Platform is spreading through England. Current figures cited by WIRED put 139 hospital trusts live and 35 of 36 integrated care boards actively using the system. One regional board has spent the past year demonstrating how slippery those verbs can become.[3]
Greater Manchester’s own adoption paper says formal onboarding requires a memorandum, two data-processing agreements, a tenant, and login credentials. The tenant can be created within a day. The same paper states that onboarding requires no active use, and that many onboarded organizations had no current plan to use the platform. NHS England was publishing onboarding statistics as a measure of program success.[1]
A login can therefore enter the national rollout dashboard before a local team moves data, changes a workflow, replaces a tool, or produces a patient benefit.
The tenant arrives before the decision
NHS England awarded a Palantir-led consortium the Federated Data Platform contract in November 2023. The expected investment reaches £330 million over seven years. Each NHS trust and integrated care board receives a local instance, with the broader project intended to standardize data models, share tools, and support work such as elective recovery, care coordination, vaccination, population health, and supply-chain management.[1]
Those are real coordination problems. Patient information still crosses hospitals, primary care, community services, spreadsheets, and paper with ugly gaps. A common technical surface could make reusable tools and national datasets easier to move. The argument for shared infrastructure has teeth.
The rollout metric blunts that argument by hiding the depth of participation. Greater Manchester’s paper separates the sequence plainly. Signing the documents creates a tenant. Supplying data populates it. Deploying a product changes work. Sustained use may produce operational value. Each stage demands different labor, governance, cost, and trust.
Compress those states into a single adoption count and the program gains a flattering denominator. Procurement success appears earlier than implementation success. A region keeping its established platform can still be listed inside the rollout once a dormant tenant exists.
Greater Manchester understood the reputational consequence. Its board paper warned that public communications about onboarding could imply deeper engagement and data transfer than had occurred. That warning came from an organization handling some of the most sensitive records in public life. Precision here is basic consent infrastructure.[1]
Manchester already built the thing it needs
Greater Manchester’s Analytics and Data Science Platform took six years to develop. In May 2025, the regional team judged its local capability ahead of the national FDP for integrated-care-board work, with some functions estimated two to three years beyond the FDP roadmap. The local system already held primary-care data, supported research infrastructure, and carried what the board described as a hard-won foundation of trust among health professionals and the public.[1]
Its red lines were concrete. Locally held GP data would stay outside the FDP. Adoption could add no unexamined local cost. The regional secure-data environment for research could suffer no damage. Public communication could not inflate the depth of engagement.[1]
NHS England answered that a bare tenant carried no direct charge, national-product consumption was covered within a fair-use cap, and local deployment would still require regional staff and governance work. Custom development and unsupported products would fall to the region. The answer confirms the basic point: free tenant creation and free implementation are different economic objects.[1]
The board ultimately recommended minimal onboarding while preserving its existing platform, monitoring the national program, and adopting FDP capabilities only when they proved superior or delivered better value. WIRED reports that the region has since hardened its refusal to use the system fully, while Palantir disputes claims that the local alternative has stronger evidence of patient or cost benefits.[3]
That disagreement deserves real evaluation. A locally built system should face the same burden of proof as a national vendor. Greater Manchester’s confidence does not establish clinical superiority by itself. Palantir’s national footprint does not establish it either. Both need outcome measures tied to specific workflows.
The metric becomes the control surface
Adoption statistics influence the February 2027 break-clause decision reported by WIRED, the credibility of a seven-year procurement, and the claim that no viable alternative exists. They also shape local bargaining power. A board recorded as active looks like a late participant asking for exceptions. A board recorded as maintaining a separate production system looks like evidence that architecture, custody, and supplier choice remain open.[3]
NHS England’s contract explainer says Palantir acts as a processor, NHS organizations control access to their instances, data stays in the UK, NHS-funded intellectual property remains with the NHS, and exit planning is built into the agreement. It also says any extension review will examine uptake, product development, benefits, strategic priorities, and the supplier market.[2]
That review needs a harder ledger. Count tenants created. Count tenants carrying live data. Name the datasets. Count deployed products and regular users. Publish workflow-level outcomes with baselines. Record local labor and integration costs. Separate capabilities built by the national program from tools retained or developed locally. Report regions that rejected a product after evaluation without laundering rejection into adoption.
The fight around Palantir attracts ideology, geopolitical anxiety, vendor advocacy, and institutional self-protection. Better measurement will not dissolve any of that. It will prevent the adoption number from quietly settling the argument on behalf of the contract.
A national health-data platform should survive honest categories. If its products improve care, the evidence will become stronger when dormant credentials disappear from the success metric. If local systems outperform it, the same categories will expose that too. Public infrastructure earns legitimacy through legible use, cost, custody, and consequence. A login proves only that somebody can log in.
Sources
[1] NHS Greater Manchester Integrated Care Board paper [2] NHS England: FDP contract explainer [3] WIRED: The Single English County Saying No to Palantir