A ministerial letter makes Palantir's NHS data platform optional for trusts

The national data opt-out cuts in one direction only. It leaves direct care alone, including live operations, and removes the record from research and planning ^[9]. That second half is the estate the federated data platform was bought to work on, so each opt-out subtracts from the asset the £330m contract exists to exploit while nothing visible changes in a clinic ^[3][9].

The size of that subtraction is not established by the record. Sixty thousand withdrawals between mid-May and mid-July works out at about 30,000 a month, or close to a thousand a day ^[2][2]. Frith calls the rise "modest" and says it will be monitored, and he attributes it to media interest in Palantir's position rather than to any measured cause ^[7][8]. A skeptic would say a minister with no denominator in front of him is describing a news cycle, not a degraded research base, and on the published evidence that objection stands: the letter gives a flow, not a stock, and no share of the eligible population. What it does establish is that the health innovation minister has now put the mistrust mechanism in writing, alongside the warning that the benefits of the 10-year health plan may not be realisable if patients stop sharing data ^[1][10].

The load-bearing change is the one that got less attention. In July a health minister told parliament that trusts would be told they "must use" parts of NHS England's federated platform; Frith now says there is no requirement to use it and that trusts may prefer an alternative ^[11]. A mandate converted into a preference moves the adoption decision to hundreds of local boards, each of which can now decline without picking a fight with the centre. Coverage of a federated platform becomes a function of local appetite, and the vendor's outcome figures become harder to extend across the system even if they hold where they were measured.

Those figures are also the part nobody can currently settle. Palantir says trusts using its technology recorded 110,000 additional operations, a 15% cut in discharge delays for long-stay patients and a 6.8% improvement in 28-day cancer diagnosis ^[5]. NHS England has reported similar numbers but says it cannot draw conclusions about cause and effect, and the government's statistics watchdog is examining the data ^[6]. So the break clause decision on a seven-year contract worth an average of roughly £47m a year will be taken before the value-for-money question is answerable, which means it will be taken on trust and politics ^[4][1].

That is the trade-off ministers actually face. Holding the contract keeps whatever operational gain exists and keeps the mistrust that Frith has now conceded in writing, along with the residue of the earlier episode in which ministers apologised after Palantir staff were found to have access to identifiable patient data ^[14]. Switching provider, which Layla Moran's committee position favours, costs migration time during which opt-outs keep accruing, and Moran has warned against deferring the choice past February ^[12]. It also runs into an unsettled legal question: Sadiq Khan blocked a £50m Metropolitan police contract on the grounds that public money should go to companies sharing the city's values, and Palantir is suing over that decision ^[13]. Whichever way February goes, the mandate is already gone, so the government has bought itself an exit at the cost of the one thing that guaranteed the platform's reach.