Every household, factory, and government agency carries a direct financial stake in whether Europe’s power grids can absorb the next industrial expansion. A proposed data centre now raises far bigger questions than simple grid access. It forces a reckoning over who foots the bill for supporting infrastructure and whether enough generation capacity exists to feed it.
Those questions grow sharper when applied to artificial intelligence. Europe needs clarity on cost allocation before committing scarce grid resources to speculative projects.
The numbers demand attention. The International Energy Agency’s 2025 base case projects global data centre electricity consumption hitting roughly 945 terawatt-hours by 2030. That more than doubles 2024 levels. Meanwhile, the agency warns that geographic clustering of these facilities complicates grid integration.
As a result, the uncomfortable economics become unavoidable. An MIT study found 95% of generative AI pilots at companies yielded zero measurable return. Separate examinations of AI data centre economics have uncovered no demonstrated profitability.
**Testing demand before building for it**
Ohio offers a compelling case study. Utility AEP Ohio reported that an initial pool of approximately 30 gigawatts in data centre requests shrank to about 13 gigawatts after customers faced mandatory engineering study fees. Binding agreements eventually covered just 5.6 gigawatts from that pool.
This illustrates why expressions of interest require financial backing before driving expensive infrastructure decisions. Forecasts gain credibility only when the submitting party accepts meaningful financial responsibility.
AEP Ohio’s data centre tariff proves practical. It demands upfront credit posting, payment for study fees, and contractual obligations to reimburse buildout costs if projects cancel or face major delays. Europe need not copy this framework wholesale. The underlying economic principle transfers regardless.
**The opportunity cost**
The larger concern involves sacrificing scarce power infrastructure for activities that have shown no business value or profitability. Europe faces a generational challenge around industrial capacity and regional security.
Manufacturing, domestic energy equipment production, and defence output generate employment across the continent. These sectors cannot lose out to AI data centre usage that has not demonstrated commensurate value.
Grid operators and policymakers should judge competing proposals by their credible contribution to employment, productive capacity, and energy security. They must weigh those factors alongside infrastructure requirements.
**Timing matters**
Building new generation capacity helps, but sequencing proves critical. Data centres become operational within two to three years. Supporting energy infrastructure requires far longer. Current wait times for natural gas turbines stretch as long as five years. Step-up transformers demand up to three years.
Solar and battery storage deploy faster. However, these resources cannot solve every constraint created by rapid load growth. Promising future generation buildout does not, by itself, resolve near-term bottlenecks.
**What regulators should require**
Regulators should mandate credible development milestones and financial commitments before incorporating large proposed loads into infrastructure plans. Developers must pay grid upgrade costs attributable to their projects. Where firm service remains unavailable, contracts should specify exactly when and how electricity use gets curtailed during system stress.
Local approvals should scrutinise water requirements, especially given recent European drought conditions. Any government subsidies should come with published public benefits and attached conditions.
Without these safeguards, customers face costs from infrastructure built around demand that never materialises. With them, planners gain firmer commitments and developers understand their obligations before proceeding. That foundation supports growth far better than treating every announced project as an urgent national necessity.
