Editorial correction — 14 October 2026: We removed unsupported causal claims, corrected the distinction between policy commitments and delivered business capacity, and identified the population and dates of the confidence survey. The original publication date is retained; this is desk research and ObvioTech analysis.
Read delivery claims with their dates attached
The government’s 29 January 2026 update reported that 38 of the AI Opportunities Action Plan’s 50 actions had been met. It described the launch of Isambard-AI in July 2025 alongside further compute investment and future expansion commitments. That is a dated account of policy delivery, not a measurement of how much usable capacity every UK business can obtain today.
Public investment, announced private investment, procurement plans and operational services are different measures. The earlier version’s shorthand about government-committed £100 billion blurred those categories. A business evaluating an opportunity should check the relevant programme’s eligibility, application route, available service and delivery date rather than treating a national investment headline as an accessible budget.
What the confidence research supports
techUK’s March 2026 polling with Public First covered 275 UK decision-makers from technology businesses and 256 from other businesses. Fieldwork ran from 24 February to 6 March, and results were weighted. In the technology group, 45% reported having considered relocating investment or operations outside the UK; 83% were exploring international expansion.
Considering relocation is not the same as moving headquarters, cancelling a UK investment or leaving the country. International expansion can also coexist with continued UK operations. The survey records reported views at a particular time; it does not establish that the AI Action Plan caused those views, nor that the respondents’ intentions subsequently became completed moves.
ObvioTech assessment: connect policy to a purchase decision
Policy progress and business confidence can move differently without either series being meaningless. Our assessment is that a company should connect the two through concrete dependencies: the service it needs, the price it can obtain, the skills available to operate it and the conditions under which it can change supplier.
For example, a national compute announcement does not resolve a team’s data-transfer requirements or prove that a particular model meets its quality target. Conversely, a challenging survey result does not establish that a specific UK service is unsuitable. Evidence at the level of the actual workflow is needed before either conclusion becomes a purchasing decision.
A short programme-checking checklist
- Record the announcement date and distinguish a funding commitment from a live service.
- Check whether your organisation and proposed use are eligible.
- Obtain the actual usage, support, location and exit terms.
- Test a small representative workload with an acceptance criterion set in advance.
- Compare the resulting costs and dependencies with a realistic alternative.
The government’s delivery dashboard is a starting point for checking programme status, rather than a substitute for those commercial checks. This article does not certify the completion of every action or forecast a return from any policy measure. Its practical conclusion is to keep national progress indicators, survey sentiment and a company’s own implementation evidence separate.
