AI Data Centre Boom: Bubble Risk and What Government Should Do Now
Scotland is being pitched as prime real estate for AI data centres. A new report from the Centre for Public Enterprise warns the surge looks like a bubble, with a real chance of stranded assets if market conditions turn.
AI's growth has been immense. According to Harvard economist Jason Furman, it accounted for virtually all U.S. economic growth in the first half of 2025. But fast growth doesn't erase fragility, and it doesn't cover the environmental and infrastructure costs that fall on the public sector.
Why this matters for government
- Grid pressure: Large data centres consume extraordinary amounts of electricity, tightening capacity and pushing up prices for households and businesses.
- Water demand: Cooling systems can place a heavy draw on local water supplies, with knock-on effects during dry spells.
- Thin job creation: After construction, facilities tend to run with small teams, limiting local economic benefits.
- Planning blind spots: Many councils don't currently require environmental impact assessments (EIA) for these projects, despite their scale and cumulative effects.
- Community impact: Noise, visual impact, and traffic are real issues if facilities are placed near homes.
There's precedent. In London's Isle of Dogs, more than 20 data centres operate, and the council has warned housebuilding could be delayed for a decade or more due to electricity constraints. That's a clear signal for Scotland's planners.
What stakeholders are saying
Campaigners warn Scotland could become collateral damage if the bubble bursts, with high energy use, minimal permanent jobs, and higher bills. Concerns include moratoriums on housing in places where data centres soak up grid headroom.
Scottish Greens MSP Ariane Burgess has flagged a lack of planning guidance and weak environmental assessments, arguing that current approaches undermine climate targets while benefiting large multinationals. For a facility to pass the "truly green" test, she points to energy efficiency, renewable sourcing, heat recovery, and protection of local ecosystems.
Industry and government positions
Apatura has proposed up to 10 AI data centres across Scotland, arguing fears over energy and water are overstated. None have yet been formally submitted for planning permission.
The Scottish Government says planning authorities must consider environmental implications and can require EIAs where effects are likely to be significant. Each case should be judged on its merits under the Environmental Impact Assessment Regulations.
The UK Government calls data centres vital to growth and public services, highlighting the AI Growth Zone in north Wales and its projection of 3,400 jobs. It says the AI Energy Council is focused on powering AI responsibly without weakening clean energy goals.
Policy actions to consider now
- Make EIAs mandatory for large data centres, with cumulative impact assessments across regions, not just single sites. See guidance on EIA from the Scottish Government: gov.scot.
- Require grid capacity and water stress tests before consent. Set clear thresholds for refusal or deferral in constrained areas.
- Mandate heat recovery to district heating where viable, with performance standards tied to planning conditions.
- Tie approvals to renewable supply agreements, demand-response participation, and peak-load caps.
- Set minimum community benefits and local job guarantees. Publish jobs-per-MW expectations and track delivery.
- Charge transparent grid and water connection fees that reflect true system costs.
- Create a siting policy: near renewables, industrial zones, and heat networks; away from housing and water-stressed catchments.
- Phase projects. Approve pilots first, then scale based on verified performance and public reporting.
- Plan for downside risk: bonds or decommissioning funds to cover stranded assets if the market turns.
- Use temporary moratoriums where housing or critical services face power constraints.
What to watch in applications
- Full disclosure of annual and peak power draw, backup fuel type, and grid connection timelines.
- Water source, consumption, discharge temperature, and drought contingency plans.
- Heat recovery design, end users, and performance guarantees.
- Jobs: construction vs. permanent roles, local procurement share, and training commitments.
- Independent environmental baselines and ongoing monitoring with public reporting.
- Alignment with national climate targets and local development plans.
Coordination across governments
Local planning teams will need clear national guidance and access to technical expertise on data centre design, grid integration, and water systems. Cross-department coordination is essential: planning, energy, climate, housing, and economic development must work off the same playbook.
For context on national energy coordination, see the UK's AI Energy Council: gov.uk.
Building internal capability
These decisions move fast. Upskilling planners, energy officers, and procurement teams on AI infrastructure and market dynamics will reduce risk and speed up sound decisions. If your department needs a concise primer or course options, explore sector-specific training here: Complete AI Training - Courses by Job.
Bottom line
AI data centres can support growth, but unchecked expansion can lock in higher bills, grid constraints, and missed climate targets. Treat the boom like a bubble that might pop. Require proof, price the externalities, protect communities, and scale only on verified performance.
Your membership also unlocks:
