If electricity and data are the ‘new oil’, is grid connectivity the strategic bottleneck in the AI transformation?

Artificial intelligence (AI) technology is advancing at an extraordinary pace. At the same time, the computing power used to train frontier AI models has been doubling every five to six months. Unlike the semiconductor scaling that defined previous technological eras, this growth is driven not by shrinking transistors but by deploying ever-larger chip clusters.

Combined with the scale of investment in data and computing infrastructure, this progress is already transforming entire industries. But the energy system must keep pace for progress to continue. This shift from silicon efficiency to physical scale is precisely why grid connectivity has become the binding constraint. Strong leadership is needed to align clean energy investments, grid build-out and AI growth.

The underlying issue is that investment in AI data centres is accelerating faster than power grids were designed to accommodate. While compute capacity, capital and talent remain critical, in many regions, connecting a new facility to the power grid can take 4-10 years, while AI data centres are typically planned and built within two to three. This misalignment increasingly determines which projects advance and which stall.

According to DNV’s Global 2025 Energy Transition Outlook, electricity demand from AI and data centres will rise sharply up to 2030, with North America consuming half of the total demand by then.

From 2035, AI training and inference (the moment where a trained model is applied to real-world data to generate answers) will become the dominant driver of data centre electricity use. By 2060, DNV estimates ~80% of data centre electricity demand will come from AI, and the sector reaches 11% (6,400 TWh) of final electricity demand, slightly less than for global space cooling demand.

Most of this additional load will connect through transmission grids. Approximately 10% of new transmission line connection requests in 2030, and 12% in 2040, will be for data centres, globally.

According to the International Energy Agency (IEA), permitting reform, grid-code harmonization, new financing models, and public-engagement efforts are moving, but not fast enough to match current AI investment. Efficiency helps, but it cannot remove the need for physical capacity, firm connections, and predictable operating envelopes.

The risk profile has shifted: access to the grid – rather than chips, capital, or algorithms – is increasingly the binding constraint. For developers, grid-connection uncertainty now rivals technology risk. For operators and policy-makers, the challenge is integrating a new demand class without compromising reliability.

Until recently, most data-centre electricity demand fell into two categories: hyperscale cloud sites (predictable, suited to long-term planning) and cryptocurrency mining (volatile but often interruptible). AI data centres sit between these extremes.

They combine very high-power density with fast, uncertain ramping and a low tolerance for interruptibility – making them the most challenging load for today’s grid planning and interconnection frameworks.

This makes AI data centres something of a stress test for grid connectivity as they turn it into a system‑level challenge. Grid operators must assess not just individual projects, but also how clusters of AI data centres interact with existing assets and each other − often with limited data and under significant time pressure.

Near term, the aim is to reduce risk and deploy systems assurance, rather than bypassing the grid. Practical measures include:

An AI campus, for example, secures capital and hardware but faces a multi‑year grid queue. By combining phased connections with on‑site storage and operational load controls, the project can begin operations earlier − while grid reinforcements are still under way − without compromising system security.

AI is already supporting grid planning and operations, such as accelerating power‑flow studies, congestion analysis, AI-enhanced digital twins. But it cannot remove the regulatory, institutional and physical risk constraints, and it will not be able to offset AI‑driven load growth within the timeframes that matter.

Leadership is needed from three stakeholder groups:

The future electrification of our societies requires a new mindset toward digitalization and systems thinking: data centres can act as an important enabler for faster scaling, thanks to their strong business case, just as we are seeing with batteries for EVs, making storage another important enabler for the energy transition. Aligning AI growth with grid connectivity is now a central leadership responsibility.