AI Data Center Power: If We Install It Home by Home, What Happens to My Electricity Bill

Ari Matusiak works on a team reshaping the US energy system. The question he opened his talk with was the audience's own bill. Open your electricity bill, he said, and don't you feel it's much higher than it used to be?

Behind that rising bill is a grid stretched taut, and one of the main sources of that tension is AI data centers. Matusiak called this arrangement — tech companies making money while consumers pay more — "a pretty ridiculous deal."

Nine out of ten American workers say their incomes have barely risen and can't keep up with the cost of living. Meanwhile, energy costs have risen 40% in the past five years.

The burden falls further down. One in three US households says paying the monthly electricity bill is hard. In parts of West Virginia, the electric bill can rival the mortgage, so heating and cooling a home costs as much as owning it.

Matusiak connected this statistic to his opening problem with the phrase "the bill I mentioned."

Matusiak said the US energy system has always had two sides. One is supply: people in hard hats and people in suits build power plants in a ritual you're not invited to. The other is demand: all citizens pay for the energy they need and share the cost of the infrastructure to produce it.

He summed it up as a deal that has held for 30 years. Prices creep up, usage stagnates, and consumers stay in a position where they pay the bill but don't participate in the system.

Matusiak said a gigawatt is a huge unit of energy and offered an analogy to grasp it.

That 270 gigawatts is enough to power two out of three US households watching Netflix all day, every day, as much as they want. AI power demand has nearly tripled in just one year, and today's system was not designed to handle that load.

If this continues, it will be one of the largest infrastructure expansions in US history, comparable to the Interstate Highway System. The problem is that the funding will be paid for by consumers' monthly electricity bills.

The money goes to infrastructure, not to consumers. Matusiak offered another analogy: the boiler is aging, the water heater is about to break, and you have to endure repair costs that weren't in your budget.

As expected, people don't like this deal. Communities across the country are resisting, and bans and protests are breaking out.

As a result, tens of billions of dollars in AI data center projects are being canceled or delayed. Yet these companies demand more power. They need it.

These companies are in a global competition they believe they must win, and they'll pay nearly any price for the energy needed to win. They need energy fast and at scale, regardless of where it comes from.

They are already investing in power plant projects, but construction takes 5–7 years. To get power faster, some companies are buying old 747 passenger jet engines and converting them into turbines that generate electricity 24 hours a day. Matusiak asked the audience to imagine the deafening noise those machines make in communities.

Matusiak says imagine these tech companies getting some of the energy they need from simple home improvements. The list includes rooftop solar panels, heat pumps, replacing air conditioners, boilers, and water heaters, batteries in garages and appliances, and software that connects with neighbors to send power back to the grid.

Nationwide, homes can supply 100 gigawatts of the 270 gigawatts data centers need. That's more than a third. By producing and storing their own energy and reducing total usage to create slack on the grid, electricity rates go down too.

The entity that should pay for those upgrades is the tech companies that need that energy right now. This is the skeleton of the exchange the talk proposes.

Matusiak suggests working through what that actually looks like at the scale of his hometown.

Imagine a tech company comes and wants to build a 200-megawatt data center. A project that size might spend $500 million, $1 billion, or more on the power it needs. But to get project approval, it will probably also negotiate a community benefits agreement — adding $10 million or $50 million to invest in community programs that benefit residents and families.

Matusiak ends this section saying that sounds pretty good, but questioning whether it solves the problem.

The industry's answer is about new energy, not fixing the grid problems residents already pay for. Whatever new generation companies build, the cost attaches to the new power, and the existing rate structure is left untouched.

So residents' electricity bills stay exactly the same. At this point, Matusiak calls the alternative a "new deal" and moves to the next step.

Imagine a notice from the utility company. It says 70% off new appliances. The old water heater gets replaced by a contractor with a new high-efficiency heat pump, costing hundreds of dollars instead of thousands.

How is that possible? Because the tech company put the $500 million, $1 billion, or more it would have invested elsewhere into residents' homes and neighbors' homes. Install solar panels, batteries, and heat pumps in 50,000 households in a city, and by Matusiak's math, that produces the same 200 megawatts of power.

This is the shape of what he calls the "new deal."

But this approach also gives tech companies what they need: power procured quickly. Retrofits can start next week and finish in 1–2 years. Compare that to 5–7 years for a power plant.

Companies receive the same credit from utilities or regulators as if they had invested in a new power plant. Moreover, because the whole investment is a community benefit, it's more likely to be accepted and built from the start. In Matusiak's words, it's a better deal for tech companies.

Residents and neighbors get equipment that is free or heavily discounted, and that equipment becomes theirs. According to what Matusiak has verified, the upgrades save up to $2,000 a year on energy bills.

The distribution method is described as a "community fund." Tech companies put money in, and residents and neighbors take discounts according to need. It's not first-come, first-served, nor for those who negotiate hardest; the households that need the most support get the biggest discounts.

The fund size is set based on the number of households, so manufacturers and contractors also agree to discounts to win a large volume of new business. Matusiak's explanation is that everyone participates together.

The fund size is set based on the number of households. So manufacturers and contractors are willing to offer discounts. They win a lot of new business from this deal. Matusiak analogized it as "like Costco, but for energy" — community energy.

A question may arise: won't heat pumps add demand to the grid? Matusiak acknowledges that sometimes they do.

But in most of the US, the grid is more overloaded in summer. So his answer is that replacing inefficient air conditioners with high-efficiency heat pumps actually reduces grid demand.

Asked whether it can work, Matusiak answers yes. There are already hundreds of home networks nationwide, organized as so-called virtual power plants.

These networks send over 30 gigawatts of power back to the grid. This is the evidence that the exchange he proposed is not new technology but an already-proven approach.

The point that software sharing home-generated power with neighbors and sending it back to the grid is the core of this structure was also mentioned earlier.

If a data center is operating or planned in your community, residents and neighbors can exert power and say, "Investing in you is the entry fee, the starting point." Matusiak lays out three methods.

First, speak up. When a neighbor talks about that data center on Nextdoor, TikTok, or any social media, reply to that post and make it a topic by informing people about community energy.

Second, don't wait for the company to bring a deal — go together and demand that this must be a deal. He added that if you don't want to go to city council meetings, send someone who does, and the audience laughed.

Third, once you get a deal, use it, get neighbors to use it too, and share what you achieved with neighbors to your network so communities nationwide can use it as a model.

Matusiak offers three reasons. First, it saves significant money on electricity bills. Second, it's better for communities and clearly better for our planet.

The final reason is having agency in the story around AI. Instead of feeling like the future just happens to us, we can have a voice in how the next chapter is written.

He defines this as "shared power, our power, power in front of everyone," and closes the talk saying each of us, our communities, and America deserve more than that.