Summary

• The Scottish government is expected to consider an SNP national council motion calling for a moratorium on new datacentres without planning permission.
• The debate follows concern that planned hyperscale projects could overwhelm Scotland’s renewable energy capacity and local planning systems.
• The clash shows UK AI strategy depending on electricity, land, grid access, and public consent as much as model development or investment announcements.

The Scottish government is facing pressure to consider a moratorium on new datacentre projects, putting a central assumption of the UK’s AI infrastructure strategy under direct political strain.

The Scottish National party’s national council has passed a motion calling for a freeze on new datacentres that have not yet received planning permission. The motion has been sent to ministers for consideration after warnings that 24 planned hyperscale projects could use more than one and a half times Scotland’s peak electricity demand.

The immediate controversy is linked to wider scrutiny of the proposed Lanarkshire AI growth zone, a major datacentre hub backed by CoreWeave and DataVita and promoted as part of the UK’s effort to build domestic AI infrastructure. Questions have been raised over whether the project can meet renewable energy claims attached to it, including whether the site has the land, grid connection, and generation pathway needed to support its power requirements.

The Scottish Parliament has also seen formal concern lodged over hyperscale datacentre plans, with a motion warning about energy demand, decarbonisation targets, local air quality, and household electricity bills. That concern points to the policy problem now facing AI infrastructure: large compute projects are being sold as strategic national assets, but their local costs arrive through planning applications, substations, grid queues, backup generation, land use, and community disruption.

Datacentres have become the physical test of AI policy. Governments can announce growth zones, sovereign AI funds, compute partnerships, and investment totals, but those commitments only become usable when facilities connect to power and networks, secure cooling, meet environmental conditions, and win planning consent. Scotland’s renewable resources have made it attractive to developers, yet renewable capacity is not the same as immediately available power for 500MW or 1GW compute campuses.

The clash also exposes the weakness of investment led storytelling. AI datacentre announcements often carry large headline figures, job claims, and national competitiveness language, while the infrastructure assumptions beneath them receive less attention. If the grid connection is uncertain, renewable generation is not yet built, or planning consent is missing, the project remains a political promise attached to a spreadsheet rather than an industrial asset.

Technology companies and cloud investors will see a Scottish moratorium as a warning that AI infrastructure cannot avoid democratic planning. Local communities are unlikely to accept enormous power consuming facilities simply because they sit under an AI label. The employment benefits of datacentres can be modest relative to their land and power footprint, and the local economic case weakens further if residents believe energy infrastructure is being diverted towards remote compute demand.

The issue is also strategic for the UK government. AI growth zones were intended to create regional clusters of power, land, compute, and investment. If Scotland begins to restrict datacentre buildout, ministers will need a more credible national plan for where compute capacity should go, how power will be allocated, and how sovereignty claims can be reconciled with reliance on foreign owned cloud and AI infrastructure.

The likely outcome is not a simple rejection of datacentres. Scotland still has universities, energy assets, connectivity, and industrial sites that could support some digital infrastructure. But the political mood is shifting from “green datacentres” as a default good to a more forensic question of which projects are technically viable, socially acceptable, and worth the energy they consume. AI strategy is now being tested at the substation and the planning committee, where rhetoric has much less room to manoeuvre.