I would like governments to stop counting impressive announcements as infrastructure. A minister can stand beside a picture of a building whenever he likes. Until the building has customers and a workable opening date, the achievement belongs primarily to the person who printed the picture.

The Guardian reports that Microsoft has pulled out of Nscale’s planned AI datacentre in Loughton, Essex, over power-supply concerns. Ministers had celebrated it as Britain’s largest sovereign AI datacentre. It was supposed to open next year. The grid operator has said it cannot supply enough electricity until the early to mid-2030s.

That is not a minor scheduling adjustment. That is enough time for the announcement to complete primary school.

Microsoft was the project’s only publicly announced customer. Nscale says it remains committed to Loughton and is exploring other power sources. Its Microsoft relationship continues elsewhere, including Norway. The site has been cleared; this is not proof the project is dead.

It is, however, a fairly good reason to stop speaking of a plan as though it has already done a day’s work.

I’m not against building useful computing infrastructure. I’m against a political economy in which describing something as enormous seems to exempt the adults involved from asking when it will function. The bigger the number, the smaller the permitted follow-up question. Eventually somebody asks about the electricity and ruins everybody’s sovereign afternoon.

A country does not become an AI superpower by possessing an excellent rendering.

The word “sovereign” is doing heroic work here. It sounds reassuringly like national control, resilience and strategic competence. None of those qualities is established by attaching the adjective to a private development whose anchor customer has left. I can call my empty spare room a sovereign research institute. The visiting professor will still need somewhere to plug in a laptop.

The Guardian first found the proposed site operating as a scaffolding yard in March. It has since been cleared. Good: physical progress. Let’s discuss that honestly, alongside the missing customer and delayed grid connection, instead of expecting a cleared site to carry the emotional weight of Britain’s economic renaissance.

What I want from ministers is ordinary, verifiable progress: financing that exists, realistic power arrangements, construction milestones and customers who still intend to arrive. Not another grand total assembled by adding together everyone’s most flattering future tense.

Nscale is entitled to pursue alternatives. Businesses change plans, solve engineering problems and sometimes succeed after setbacks. Government should be able to describe that uncertainty without acting as if a caveat will personally assassinate growth.

There are people capable of building this stuff. Give them a grid strategy, serious scrutiny and time to do the work. Stop demanding that the ribbon be cut before anyone has established what it’s attached to.

For now, Britain has produced an ambitious proposal, a major customer’s withdrawal and another lesson in the difference between national capacity and national publicity.

The announcement is fully operational. Someone should check on the datacentre.