Five police forces in the East Midlands are ending a two-year pilot of Palantir's Foundry platform and will not take the project forward.
The East Midlands Special Operations Unit said the decision followed scoping work, consideration of funding streams and an assessment of what the project would deliver.
The available reporting does not establish that the technology failed. It does show that a long-running pilot did not produce a sufficiently funded and defined path to scaled purchasing.
A pilot is a commercial decision, not a product demonstration
Project Nectar explored data federation, AI in policing and advanced analytics. Home Office funding covered the East Midlands pilot, although the costs were not disclosed.
This is a familiar enterprise-technology pattern. A supplier proves that a platform can connect data or produce useful analysis, but the buying organisation has not agreed the outcome, operating model, ongoing cost or route into production.
By the end of the pilot, technical possibility can be clear while the commercial case remains unresolved.
What should have been decided at the start
Before a strategic technology pilot begins, procurement should require:
A defined operational outcome and baseline against which improvement will be measured.
A named owner for adoption, data governance and benefits realisation.
A total cost for scaled deployment, including integrations, support and internal resource.
Exit criteria and a plan for extracting data and learned configurations.
A competitive route to production that does not give the pilot supplier an unfair structural advantage.
A decision date and evidence pack for approving, changing or ending the programme.
A cheap or externally funded pilot is not low risk if it creates dependency before the buyer has tested the long-term commercial model.
Do not overread the result
Seven forces in eastern England have reportedly used a separate rolling Palantir pilot, and Palantir continues to hold other significant UK public-sector work. The East Midlands decision is not evidence that every deployment lacks value.
It is evidence that a buyer can complete substantial scoping and still decide not to proceed. Ending a pilot is a legitimate outcome when funding, value or delivery conditions do not justify scale.
The bottom line
Innovation programmes often count pilots launched. Procurement should count decisions reached, evidence created and weak cases stopped before dependency grows.
A successful pilot is not one that proves the software works. It is one that gives the organisation enough evidence to scale, redesign or walk away.
Sources
Liberty Investigates: Five police forces to drop Palantir tool, 19 September 2026.
Financial Times: Five UK police forces end Palantir project, 21 September 2026.