Artificial intelligence runs on computer chips, but those chips run on electricity, and lots of it. As tech companies race to build ever larger data centres, the availability of power has become one of the biggest limits on how fast the AI boom can grow, and one of the most important factors for investors to understand.
Why data centres need so much power
A data centre is a building full of servers that store and process information. AI data centres pack in powerful chips that run hot and work almost continuously, so they need enormous amounts of electricity both to operate the chips and to cool them. A single large AI site can use as much power as a small city.
Our guide to data centre stocks explains the businesses that build and run these facilities.
Where the bottleneck is
The problem is not only generating enough electricity but getting it to the right place. Connecting a new data centre to the grid can take years, because new transmission lines, substations and local network upgrades all need planning permission and construction. In some regions, the queue of projects waiting for a grid connection stretches well into the next decade.
Power companies also have to plan for demand they cannot be sure of. Building new power stations is expensive and slow, and utilities and regulators are wary of loading costs onto ordinary households for capacity that might not be needed if the AI boom slows.
Local opposition and politics
Data centres have also become a political issue in some areas. Residents worry about rising electricity bills, water use for cooling, noise and the loss of land. That has led to stricter local rules, delays and, in some places, outright bans. When political support for new building weakens, it can slow the pipeline of projects and change the outlook for companies that depend on them.
How companies are responding
Tech giants are trying several routes around the bottleneck. Some are signing long-term contracts with power producers, including nuclear plants. Others are building their own generation on site or investing in new technologies such as small modular reactors. Many are looking for locations with spare grid capacity, even if they are far from big cities.
All of this adds to the cost of building AI infrastructure, which is already very high. Our explainer on AI capital spending looks at why the size of that bill matters to markets.
Who it affects in markets
Power constraints ripple through several sectors. Utilities with spare capacity or a strong pipeline of new generation can benefit. Makers of electrical equipment, such as transformers, cables and cooling systems, have seen demand surge. On the other side, chipmakers and cloud companies can see their growth limited if customers cannot get the power to switch on new servers.
Energy prices matter too. Strong demand from data centres can push up wholesale electricity costs, which feeds into inflation and household bills.
The beginner takeaway
The AI boom is not only about software and chips. It depends on physical infrastructure, and electricity is one of the hardest parts to scale quickly. When you read about AI spending plans, it is worth asking where the power will come from, because that question increasingly shapes which companies win and how fast the industry can grow.
Seeing the links between sectors is what turns headlines into a real view of the market. If you want an honest view of where your own trading stands, take our free Trader Assessment and see where your strengths and gaps really are.
