The Industry Powering AI: Why the Next Great Engineering Project Is the Power Grid

The data driving artificial intelligence is also redrawing Europe's industrial energy map

Last updated: September 2026

We talk a lot about algorithms, but artificial intelligence is, above all, an infrastructure problem. According to the International Energy Agency’s (IEA) latest report, electricity demand from data centres surged 17% in 2025, while demand from AI-focused data centres specifically grew by 50% over the same period. The projection for 2030 is even more striking: from 485 TWh to 950 TWh globally, nearly doubling in five years. Just five major tech companies invested over $400 billion in 2025 alone — a figure that already exceeds global investment in oil and gas production.

This isn’t a minor detail for the industrial sector: it means AI has effectively become a first-tier industrial consumer, with the same grid, power, and planning demands as a steel mill or a chemical plant.

When a Data Centre Behaves Like a Factory

The “From Chips to Grids” report (Interface, May 2026) puts it plainly: large AI compute clusters — so-called “AI gigafactories” — behave like “electro-intensive industrial plants connected to already strained grids.” A single cluster can concentrate demand of up to 300 MW at one site. It’s no coincidence that OpenAI paused its UK data centre project in April 2026, citing electricity costs and regulatory uncertainty.

The lesson for industrial engineering is direct: energy availability, not just compute capacity, is becoming the real bottleneck of technological growth.

Digitalising the Grid as the Answer

The good news is that the same technology creating the problem also offers part of the solution. The European Commission’s roadmap for energy system digitalisation (June 2026) estimates that grid digitalisation could generate €71 billion a year in direct savings for consumers, plus over €300 billion in systemic benefits — fewer losses, better renewable integration, and predictive maintenance of critical infrastructure.

At the same time, 92% of manufacturers surveyed in recent industry studies name smart manufacturing — sensorisation, grid digital twins, real-time analytics — as the main driver of competitiveness over the next three years. Industrial sustainability is no longer just about plant-level efficiency; it’s about how that plant interacts with an increasingly complex grid now shared with massive new consumers like AI.

A Closing Thought

AI’s energy challenge is, at its core, a systems engineering challenge: designing grids, plants, and processes capable of absorbing demand that will double in less than five years — without sacrificing decarbonisation goals. Organisations that grasp this early, integrating consumption data, predictive maintenance, and energy planning from the design phase, won’t just be better prepared for the transition; they’ll turn a constraint into competitive advantage.

How is your organisation preparing for industrial electricity demand that keeps climbing? We’d love to hear your perspective in the comments.

Explore similar insights