Scottish AI datacentre project exposed over unrealistic renewables pledge

The Lanarkshire story is a microcosm of the physical constraints now biting the AI boom. On paper, a fully renewables‑powered £8.2bn AI growth zone, anchored by CoreWeave and DataVita, ticks every political box: green jobs, digital infrastructure, and a narrative that AI can be decarbonised by clever siting. The Guardian’s FOI‑driven reporting, however, suggests that the promise of 1GW of new local energy was never viable and that the project will quietly lean on an already strained UK grid.

For the AGI race, this matters because frontier training and inference are concentrating in exactly these hyperscale datacentres. If local communities and grid operators conclude that the numbers don’t add up—on land use, water, or renewables—permits will slow and costs will rise. That could drag out timelines for scaling compute, particularly in politically sensitive democracies, even as demand for model capacity explodes.

It also undercuts the “AI growth zone” rhetoric emerging in several countries. If flagship projects are seen as over‑promising on jobs and sustainability, public support for AI infrastructure will erode, making it harder for labs to secure social licence for the massive energy footprint that AGI‑class training runs will require.