Microsoft CEO Satya Nadella dropped a number at Davos this week that should matter to anyone with an electricity bill: AI data centers are projected to consume 10-12% of total U.S. electricity by 2030.

Let that sink in. One in every ten kilowatt-hours powering America - not for hospitals, not for homes, but for training language models and generating images of cats wearing cowboy hats.

Nadella framed this as a competition: whichever country can produce the cheapest energy wins the AI race. That's probably true if you're running Microsoft. But if you're a regular person trying to keep the lights on, the question isn't who wins - it's who pays.

Here's how this plays out. AI companies are building massive data centers that require billions of dollars in infrastructure investment and gulp electricity at a scale most people can't comprehend. A single large language model training run can use as much power as a small city for months.

Right now, tech companies are funding their own power generation - buying up retired coal plants, investing in nuclear reactors, and cutting deals with utilities for dedicated capacity. But energy is a scarce resource, and when demand goes up, prices follow.

The infrastructure bill for AI isn't coming from tech companies alone. It's coming from ratepayers, taxpayers, and anyone connected to the grid. Utilities will need to build out transmission capacity, upgrade substations, and secure baseload power. That doesn't happen for free, and it doesn't happen fast.

So what does this mean for your wallet? If AI is consuming 10-12% of U.S. electricity by 2030, that's demand getting pulled from the same grid that powers your house. Utilities operate as regulated monopolies - when their costs go up, they pass them along. Your electricity bill doesn't care whether the power is going to a ChatGPT server or your refrigerator.

There's also a stock market angle here that nobody's talking about. Energy infrastructure is about to become one of the most important bottlenecks in tech. Microsoft, Google, Amazon, and Meta are all racing to build AI capacity, but they can't just plug in a warehouse full of GPUs and call it a day. They need power, cooling, and grid access - and all of that takes years to build.

Utility stocks have historically been boring, low-growth dividend plays. That calculus might be changing. If AI is driving 10-12% of electricity demand, the companies supplying that power are going to see a revenue surge. The question is whether they can build fast enough to meet it, or whether energy costs become the limiting factor that slows down the entire AI boom.