Goldman Sachs projects global data center power demand will rise 165% by 2030 — a surge no existing electricity grid was designed to absorb.1
The scale of AI's energy appetite is stark. A single ChatGPT query consumes roughly 10x the electricity of a Google search.2 Training next-generation large language models requires power equivalent to a small city.2 Microsoft, Amazon, and Alphabet now rank among the largest electricity consumers on the US grid.2
The mismatch is global. Most national grids were engineered for 1-2% annual demand growth.2 The US, Europe, and Asia all face the same constraint: legacy transmission infrastructure cannot absorb an AI-driven demand spike at this speed. In Europe, grid congestion is already pushing data center operators out of Amsterdam and Dublin into lower-density markets. Singapore has imposed capacity moratoriums due to power and land limitations.
Utilities near hyperscaler campuses hold structural pricing leverage worldwide. Power purchase agreements with AI operators are effectively price-inelastic — hyperscalers cannot tolerate outages the way industrial customers can. That asymmetry shifts negotiating power toward the generator.
Nuclear operators carry a particular global advantage. AI companies have designated 24/7 carbon-free power as a procurement requirement, not a preference. Nuclear baseload meets that specification. Intermittent renewables require storage additions that raise effective costs. France, with its nuclear-heavy grid, is emerging as a preferred data center destination for European operators.
In the US, Bitzero Holdings (AIBZ) is among the operators positioning for the AI energy infrastructure opportunity.2 Globally, capital is flowing toward any generator with contracted capacity near major data center hubs — from Northern Virginia to Frankfurt to Tokyo.
The logic is simple: AI energy demand is not discretionary. Hyperscalers have committed multi-year capital programs totaling tens of billions in data center construction. The electricity to run those facilities must come from somewhere. Operators with the right geography, grid interconnection, and generation mix are positioned to capture that demand before interconnection queues tighten further.
For market participants globally, the question is no longer whether AI creates utility upside. It is which operators — in which countries — have the grid position and permitting status to capitalize.
Sources:
1 Goldman Sachs Research, Data Center Power Demand Forecast, 2026
2 Via News Signal Analysis — AI Energy Infrastructure Hypothesis, June 23, 2026
