Here’s a fact that surprises people: some of the biggest constraints on AI today have nothing to do with chips. They have to do with electricity — and the boring, physical business of getting it to a building.
Data centers are power plants in disguise
A large AI data center can draw as much power as a small city. To run one, you don’t just need chips — you need a connection to the grid big enough to feed them, transformers to step the power down, and cooling to carry the heat away. Each of those can take years to secure.
Why this is a real limit
Grid connections and large transformers are in short supply, and you can’t conjure them overnight. So a company can have all the chips in the world and still be stuck waiting on a substation. That’s why “where can we get power?” has quietly become one of the most important questions in AI.
How to read it
When power is the bottleneck, the winners shift — toward whoever controls energy, grid gear and cooling. THE ENTITY tracks the energy layer alongside the chips, so you see the whole picture, not just the part that makes headlines.
See it live on the map
Plans from $19/mo — cancel anytime. Two minutes to your first read.