← Insights
Sector studySector study · 4 chapters · 1 chart·For CIO, CISO, Head of Operations, Chief Risk Officer

Your Biggest New Customer Is Also Your Biggest Grid Risk

AI data-centre demand will more than double to ~945 TWh by 2030 (IEA), straining the grid that AI is also the best tool to manage. Energy and compute have merged into one optimization problem. A field guide for utility CIOs and heads of operations.

Data-centre electricity demand by 2030, more than Japan (IEA)
945 TWh
Transmission capacity AI could unlock with no new lines (IEA)
175 GW
Energy workers near retirement per entrant under 25 (IEA)
2.4 : 1
Chapter 01

AI flipped from feature to demand

For a decade AI helped the energy system. In 2026 it became the largest new source of demand on it, and the only tool fast enough to manage the volatility it creates.

The IEA projects data-centre electricity use more than doubling to around 945 TWh by 2030, more than Japan consumes today, with data centres driving roughly a tenth of global electricity demand growth, and over a fifth in advanced economies. AI is now the grid's biggest new customer. And it is simultaneously the grid's best tool: the same IEA analysis finds AI-enabled grid management could unlock up to 175 GW of transmission capacity with no new lines.

The two have merged into a single optimization problem. The clearest sign of the future is the first commercial-scale, power-flexible AI factory, which modulates its load in response to grid signals, turning the biggest threat to the grid into a balancing asset.


Chapter 02

The boring layer decides who wins

Grid, asset, and market value is locked behind unintegrated operational data, and that integration is where the work actually is.

Predictive maintenance, grid optimization, and trading all depend on merging operational-technology data, SCADA, historians, GIS, with IT data like market prices and forecasts. Done well, that creates a complete operating picture of the value chain; done poorly, AI stays a pilot. As the chart makes plain, the OT/IT integration layer is the enabler beneath every operational use case; it is the least glamorous and most decisive part of the programme.

There is a human constraint stacked on the technical one. The IEA finds utilities and oil & gas carry roughly 40% lower concentration of AI-skilled workers than tech, with 2.4 energy workers near retirement for every entrant under 25, a talent cliff no AI roadmap survives without a reskilling plan.

Figure · Energy AI: where the value and the constraint sit (modelled)
OT/IT data integration (enabler)
90
Grid optimization & balancing
88
Predictive maintenance (OT data)
80
Energy trading & forecasting
72
Corporate productivity
30

A Rindogatan-modelled index; the integrated OT/IT data layer is the enabler beneath every operational use case. Directional, not survey data.


Chapter 03

Resilience makes it sovereign

For critical infrastructure, control of the AI dependency is part of the resilience case, and EU spend is already moving.

Grid and trading data is sensitive in a way that goes beyond privacy: it is a matter of resilience and security. Under NIS2 and the Critical Entities Resilience directive, an operator is accountable for the AI supply chain, and a critical workload mediated by a model in another jurisdiction is exactly the dependency the sector exists to engineer out. Gartner expects Europe's sovereign-cloud spending to grow to $12.6B in 2026, naming energy and utilities among the core buyers, the market is already voting.

Sovereignty here is best framed as control and resilience rather than data residency alone: bring the inference inside the controlled, audited perimeter the operator already maintains. The figures are the IEA's and Gartner's; the modelled index is ours.

The same data centre that strains a substation can now bid flexibility back into the market. Energy and compute have merged into one optimization problem, and the binding constraint is the integrated data layer underneath the grid.


Chapter 04

Integrate the data, own the dependency

Four moves to make energy AI as resilient as the infrastructure it serves.

Treat OT/IT data integration as the programme, not a precursor to it, the operating picture it creates is what every operational use case depends on. Classify operational and trading workloads as sovereign-by-default and map them to NIS2/CER obligations.

Stand up sovereign inference inside the operational perimeter, keep the frontier to the non-critical edge, and pair every deployment with a reskilling plan for an ageing, AI-thin workforce. For critical infrastructure, controlling the dependency is the requirement; sovereignty is how you meet it.


Sources & methodology
  • 1. Headline figures are Rindogatan models, directional benchmarks to be calibrated to a specific institution, not survey statistics.
  • 2. Partner data points are drawn from publicly published research (e.g. Snowflake's Modern Marketing Data Stack, Databricks' State of Data + AI) and cited for direction only.
  • 3. Regulatory references: EU AI Act, Reg. (EU) 2024/1689; GDPR, Reg. (EU) 2016/679; DORA, Reg. (EU) 2022/2554; NIS2, Dir. (EU) 2022/2555.
  • 4. Sovereign deployment modelled on European sovereign infrastructure.