A 1,000-year-old Scottish castle faces demolition for a £5 billion hyperscale data centre. This draft examines the engineering trade-offs, heritage constraints, and operational realities for practitioners facing similar siting decisions. By Alasdair Ferguson.

In Auchtertool, Fife, a 1,000-year-old castle sits in the path of a proposed £5 billion, 600MW hyperscale data centre. Historic Environment Scotland (HES) refused protected status for Hallyards Castle, citing that planning proposals were “at an advanced stage.” Residents argue the listing application arrived shortly after ILI Group’s initial planning-in-principle submission, suggesting the process was rushed. For engineers, the core tension is between the massive compute capacity required for modern AI workloads and the immutable constraints of site history.

The technical narrative is clear: ILI Group plans a 600MW facility, a significant draw on the local grid. The developer claims the castle will be “fully excavated and recorded by qualified archaeologists” before demolition, with stonework dismantled and reused in a publicly accessible setting. This is not preservation by stasis, but a managed archaeological extraction. Practitioners must ask whether the resulting data centre’s operational lifespan justifies the permanent loss of a medieval bishop’s residence. The source notes HES acknowledged the castle is “of national importance,” yet chose non-designation.

Operational constraints are significant. The proximity of a 25-hectare industrial facility to 200 homes raises standard zoning and environmental impact questions, yet the source indicates a detailed Environmental Impact Assessment (EIA) has not yet been submitted. For a working engineer, the decision hinges on whether the AI service’s performance gains outweigh the reputational and regulatory risk of building on a protected site. The recent 12-month MSP pause on data centre decisions in Scotland further complicates timelines, though HES stated the pause “does not alter the development proposals,” which remain “at an advanced stage.”

[Read More]

Tags data-and-analytics software-engineering cloud-and-infrastructure