AI and Electricity: Data Center Demand Set to Nearly Double by 2030, the Grid Isn't Keeping Pace

According to International Energy Agency (IEA) data cited in early August 2026 by analysts at Century Financial and Zentara Solutions, global energy investment is projected to reach $3.4 trillion in 2026. Of that total, $1.6 trillion is earmarked for electricity (generation and infrastructure), but transmission and distribution grids — the lines, substations and transformers that physically deliver power — capture only $550 billion, roughly 16% of total energy investment and a third of electricity-specific investment.

At the same time, global data center electricity consumption is projected to rise from 485 TWh in 2025 to 950 TWh in 2030 according to the same sources — a 96% increase in five years, driven by AI training and inference workloads. The growth rate of demand is significantly outpacing investment in the infrastructure meant to deliver it.

This mismatch echoes a mechanism documented by economist Albert Hirschman in his theory of unbalanced growth (1958): capital flows preferentially toward directly productive activities — here, compute facilities and generation capacity — ahead of the social overhead capital that makes them actually deliverable, such as transmission networks. The imbalance eventually self-corrects, but with a lag that exposes users to interconnection queues, congestion and price spikes during the catch-up phase.

For the sectors Cardan-AI tracks, the operational takeaway is direct: electricity availability and cost are becoming a siting and capacity-planning criterion as structural as chip or cloud access — particularly for industrial, aerospace and energy sites that are electrifying their own processes alongside their AI workloads.