Britain Plans Backwards — Then Sends You the Bill

Why infrastructure costs balloon when ministers announce first and engineer later

Shane Oxer — Campaigner for fairer and affordable energy

Britain’s infrastructure costs do not “unexpectedly” rise. The system makes them rise.

The routine is now familiar: announce the target, choose the site, publish an attractive price — and only then ask whether the grid, water, heat network, transport links, skilled workforce and supply chain actually exist.

A dependency found before commitment is a design decision. Found afterwards, it becomes five separate bills:

– Redesign: drawings, permissions and contracts must be changed.
– Delay: people, land and equipment are paid for while nothing useful is delivered.
– Finance: interest accumulates before the asset earns or saves a penny.
– Risk: contractors price uncertainty into every bid.
– Duplication: missing systems are bolted on separately instead of being designed once.

This is how political optimism becomes commercial risk — and how commercial risk becomes a public bill.

The Government effectively admitted this in its “2026 review of mega-projects” (https://www.gov.uk/government/publications/value-for-money-vfm-study-on-the-governance-and-budgeting-arrangements-for-mega-projects/value-for-money-study-governance-and-budgeting-arrangements-for-mega-projects). It found that ministers and senior officials are incentivised to begin too much construction before designs and delivery plans are mature. Costs and schedules are fixed too early, then projects require expensive delays, resets, scope cuts or cancellation.

That is not bad luck. It is planning backwards.

Take data centres. Scotland is selling “green compute” on the strength of renewable electricity, while its “own AI strategy” (https://www.gov.scot/publications/scotlands-ai-strategy-2026-2031/pages/8/) says that only by 2027 will it identify heat-offtake opportunities and publish guidance on what makes a data centre “green”.

Promote the investment first. Define “green” later. Find a customer for the heat later still.

Renewable electricity alone does not make a data centre green or cheap. Build it far from heat demand and its waste heat may have no buyer. The operator still pays to remove that heat, households still pay for separate heating, and someone still pays for any grid reinforcement. One energy input creates two paid-for systems because nobody joined them up.

Norway puts the engineering question before construction. Since April 2025, “new data centres above 2 MW” (https://www.regjeringen.no/en/documents/the-data-centre-industry-a-sustainable-industry-of-the-future-for-the-digital-norway/id3112356/?ch=5) have required an approved cost-benefit analysis of waste-heat use before building begins. In Oslo, Skygard’s heat could serve up to 12,000 flats; another centre at Ulven supplies heat for 5,000 homes.

Finland followed the same logic. “Microsoft’s sites at Espoo and Kirkkonummi” (https://www.fortum.com/services/heating-cooling/data-centres-helsinki-region) were chosen partly for their proximity to Fortum’s district-heating network. The scheme expects to reuse about 75 per cent of the centres’ waste heat annually and eventually provide around 40 per cent of the area’s district heat.

The heat customer helped determine the location. It was not invented afterwards.

Solar power exposes the same mistake. A field may be the cheapest site for a developer while becoming a costly choice for the country once grid works, curtailment, storage, access, lost agricultural output and restoration are counted.

France “defines agrivoltaics in law” (https://www.legifrance.gouv.fr/jorf/id/JORFTEXT000049386027): farming must remain the main activity; crop output must normally stay above 90 per cent of a control plot or accepted reference; agricultural income must be maintained; and land made unworkable by the installation is limited.

The Netherlands uses a ““solar ladder”” (https://www.rvo.nl/onderwerpen/zonne-energie/beleid): buildings first, developed land next and countryside after that, with preference given to combining uses. These countries put the hidden costs into the planning decision before consent. Britain too often discovers them afterwards — or leaves somebody else to carry them.

Heat planning tells the same story. “Denmark passed its Heat Supply Act in 1979” (https://ens.dk/media/6433/download) and completed municipal heat plans by 1990. Its method starts with demand, then compares technologies and long-term costs to find the lowest heat price for consumers.

England is only now “progressing its first heat-network zones” (https://www.gov.uk/government/collections/heat-network-zoning), ahead of the zoning legislation coming into force. Every estate, industrial site or data centre built without a heat plan makes the eventual network harder and dearer. Retrofitting is planning’s compound interest.

Then there is the electricity grid. Great Britain allowed a “connection queue of more than 700 GW” (https://www.neso.energy/industry-information/connections-reform/connections-reform-results) to accumulate — four times what is needed by 2030 — while ready projects waited as long as ten years. The old first-come, first-served system created delays, wasted resources and, in NESO’s words, sometimes unnecessary upgrades paid for by consumers.

Only now is the queue being organised around what is ready, needed and in the right place. The plan arrived after the gridlock.

Switzerland applied the correct order to rail. “Rail 2000” (https://www.alptransit-portal.ch/en/overview/construction/events/ereignis/rail-2000/true) began with the service: a coordinated timetable with reliable connections. Infrastructure and trains were then targeted at the bottlenecks preventing that service. Britain too often announces the project and price first, then redesigns the railway around whatever survives.

Britain did not lead the Industrial Revolution by announcing targets. It surveyed, standardised, designed and built. That era was neither clean nor fair, but it understood something today’s political system has forgotten: engineering is not decoration added after the decision. It is how a sound decision is made.

Politics should choose the outcome. Engineers should define the dependencies, whole-system cost range, operating model and stop-go tests before a promise becomes policy.

Costs do not balloon because concrete suddenly becomes mysterious. They balloon because uncertainty is carried forward until it becomes contractual. Contractors charge for it. Lenders collect interest on it. Consultants manage it. Lawyers renegotiate it. Taxpayers and bill-payers absorb it.

Britain’s engineers have not forgotten how to build. Government has forgotten that the cheapest time to listen to them is before the press release.