Skip to main content

​
What is the argument for green AI compute in Scotland?

Scotland should invest in green, AI-focused compute as a piece of national economic infrastructure. The global AI economy is now constrained by compute — which requires reliable electricity, effective cooling, and planning permission at scale. Scotland has abundant renewable energy (including massive curtailed wind), a cool climate for efficient cooling, and digital connectivity. These are precisely the conditions modern AI infrastructure requires. The paper argues this is a strategic choice Scotland should make now, while the window exists.

​
What does success look like?

A Scotland where AI companies are founded, scaled, and headquartered not because of short-term incentives, but because this is where AI work is genuinely possible. Where training models and deploying systems does not conflict with climate commitments because the underlying compute infrastructure was designed around clean power from the outset. Where talent stays and circulates between companies, building compounding local knowledge.

​
What global constraint is driving this?

AI progress is now limited by compute, not ideas. Grid congestion, rising energy costs, and planning complexity determine where AI infrastructure can be built. As compute becomes scarcer, geography matters again. Regions that can host AI infrastructure responsibly will shape where companies form and where economic value accumulates. The scale of wasted energy makes the case urgent. In 2025, 10 TWh of wind energy was curtailed in Great Britain at a total cost of £1.5 billion (£343 million in payments to wind farms plus £1.1 billion in replacement gas generation) — up 22% on the previous year. 98% of that curtailed energy comes from Scotland. Constraint costs are forecast to reach £4-8 billion annually by 2030.

​
How does this create a startup ecosystem?

Infrastructure alone does not create an ecosystem. A time-bound compute offer for Scotland-based startups lowers the single biggest barrier to building AI products. As companies scale from experimentation to production, they transition from supported access to commercial use. Talent stays, people move between companies, and informal knowledge flows through repeated collaboration. This is how durable technology centres form.

​
Why does Scotland fit this role?

Scotland generated 113 per cent of its electricity consumption from renewables in 2022, with 15.5 GW of operational renewable capacity and 26+ GW in the development pipeline. Its climate supports efficient cooling. Its digital infrastructure connects it globally. The University of Edinburgh has over 60 years of AI research heritage. The EPCC national supercomputer (£750 million confirmed, operational early 2027) will provide sovereign compute for research and industry. The demand signal is clear: 17 hyperscale data centre applications are currently in Scotland’s planning system, with a combined capacity of 4.4-4.9 GW — larger than Scotland’s 4 GW winter peak demand. Grid investment of £32 billion+ is underway (SSEN Transmission £22 billion for north Scotland, SP Energy Networks £10.6 billion for central and southern Scotland). The building blocks exist — what remains is a deliberate choice to assemble them.

​
Context

This article was co-authored with Joel Cohen and produced through the Building Scotland project. It connects to themes raised by Chris Herd on Scotland as a “maker” nation and feeds into the Scotland in the AI age policy whitepaper.

​
Published content