A £350,000 Prototype Contract in Search of a Buyer

Analysis as of 16 August 2026.

At noon on 3 August 2026, UK Defence Innovation reopened its Security Open Call: a two-stage competition offering fixed-price research contracts of up to £350,000 before VAT, placed by the defence secretary through Dstl, against twenty-three challenge areas defined by the Home Office and the National Protective Security Authority. Bids close on 27 August; second-stage decisions land on 14 December, with contracts expected roughly eight weeks later. A funded project runs up to twelve months to a relevant-environment demonstration, or eighteen months if an optional operational trial is exercised — a trial the sponsor states it has no duty to trigger, and one that must fit inside the same ceiling rather than adding to it.

The arithmetic around that ceiling is more interesting than the ceiling itself. Grossed up at 20 per cent VAT, £350,000 becomes £420,000 — a mere £4,560 above the £415,440 defence-and-security procurement threshold in force since January, and more than triple the £135,018 figure that applies to ordinary central-government purchases. The draft contract carries aggregate liability caps of £5 million, roughly fourteen times the maximum price. Payment comes strictly in arrears against deliverables, with no pre-funding, at most six interim claims, and a final instalment of at least a fifth of the total — meaning £70,000 or more can hinge on satisfactory completion. For an early-stage company, the instrument is contracted research revenue with real balance-sheet demands, not a grant.

What the paperwork never supplies is the customer. The bodies that wrote the challenges are not the body signing the contract, and neither is the body that would eventually buy: a police force, a Home Office command, an infrastructure operator, a venue. UKDI itself cites a ringfenced budget of at least £400 million a year, and policing reform promises a £141 million technology package — £115 million over three years for AI adoption plus £26 million for facial recognition — but nothing published reserves any of it for what this call produces. The three planned 2027 cycles are themselves conditional on stakeholder funding that has not been confirmed.

The receiving system's track record quantifies the gap. The National Audit Office found that body-worn cameras took thirteen years to reach every force, that an evidenced video-response tool was in use by just six forces, and that of 922 innovations catalogued by the College of Policing between May 2024 and September 2025, only 22 were recommended for sector-wide adoption — about one in forty. Forces spend around £2 billion a year on technology, 97 per cent of it maintaining legacy systems. The predecessor accelerator's own showcase cases took five years and three funded engagements to reach product launch in one instance, and seven years and four engagements to reach a first recorded customer purchase in the other.

None of this proves the design is wrong. Committing production money at prototype stage could lock the state into immature kit, and procurement law already permits an end-to-end competitive route from research through to final tender — a route this call, on its published terms, has simply not taken. The competition ends at a demonstration; everything after it is described as possible, discretionary and subject to another contract.

Three tests, none of them yet passed on the public record, will show whether the reopened call is a pathway or a treadmill:

— Will any Cycle 1 purchase order, or a subsequent tender notice, name the requirement owner, the acceptance authority and the budget line for an operational purchase before the research work ends?

— When the optional trials are exercised, will the evidence protocols be agreed in advance with the body that would actually deploy the system — or generated in substitute environments a later buyer must pay to repeat?

— And across the cohort, what fraction of completed demonstrators will progress within, say, two funding cycles to a paid trial, a framework position or a contract award, against a system whose measured conversion rate today is one in forty and whose adoption clock runs in years?