Skip to main content

​
What this paper is

A draft policy whitepaper produced through the Building Scotland project in collaboration with industry partners, prepared for the Scottish ecosystem in early 2026. It argues that Scotland has a narrow window to establish itself as a significant player in the global AI economy rather than becoming a dependent consumer of technologies built elsewhere.

​
What is the strategic case?

The global AI economy is no longer defined primarily by algorithmic innovation. It is defined by access to compute power, energy, and data. An intelligence arms race is underway, led by the United States and China. It is neither practical nor proposed that Scotland build frontier foundation models. The opportunity is in smaller, task-specific models and in creating the conditions — infrastructure, skills, data access, and market demand — for companies to build here. Scotland’s advantages are specific: massive renewable energy surplus (98 per cent of Great Britain’s curtailed wind energy comes from Scotland), world-class AI research institutions anchored by Edinburgh’s 60-year heritage, distinctive public-sector data assets through NHS Scotland and other bodies, and a confirmed national supercomputer at EPCC.

​
What are the five strategic foundations?

Sovereign compute infrastructure. Co-locate data centres with wind farms to absorb curtailed energy. Seventeen hyperscale applications totalling 4.4 to 4.9 GW are already in Scotland’s planning system. Provide time-bound compute access for startups to lower the biggest barrier to building AI products. Intelligence base and model access. Secure access to open foundation models through Horizon Europe partnerships. Invest in domain-specific models for Scottish priority sectors. Reduce dependency on US and Chinese providers. Data assets and ethical access. Unlock Scotland’s distinctive data through synthetic data services and data cooperatives. Enable innovation while protecting individuals. Pilot in health data through existing Safe Haven infrastructure. Skills and talent at all levels. Retain frontier AI talent through compute access and ecosystem support. Build broad workforce AI literacy so citizens thrive alongside AI. Lead with a narrative of augmentation and empowerment. Market creation through public-sector action. Use procurement, regulation, and convening power to create demand for Scottish AI solutions. Reform procurement to include simplified fast-track for SMEs. Develop a Scottish AI Quality Mark aligned with emerging standards.

​
What barriers does the paper acknowledge?

The paper is candid about risks: grid capacity constraints, environmental concerns about data-centre proliferation, devolved-versus-reserved governance complexity, fiscal pressures, and limited delivery capacity across agencies. It recommends a coordinating taskforce as an immediate priority.

​
How are the recommendations structured?

Recommendations are phased across three horizons. Immediate priorities within 12 months include establishing the taskforce, a strategic data-centre framework, compute vouchers, and procurement reform. Medium-term actions at 12 to 36 months include a synthetic data service, task-specific model funding, scaled AI literacy, and a quality mark. Longer-term structural investments at three to five years include addressing the Series A/B funding gap, rural data economy participation, and regulatory sandboxes.

​
Context

This whitepaper synthesises themes from the full Building Scotland conversation series — particularly Colin Carmichael on Scotland’s AI window, Chris Herd on the maker-or-taker choice, Nick Sherrard on system accountability, and Vicky Brock on infrastructure-first thinking. The green AI compute article provides a focused argument on the compute foundation.

​
Published content