By Matthew X. Roling
The roughly 5,500 data centers currently operating in the United States require about 176 terawatt-hours of electricity a year, which is about the same amount of power used by 16.7 million households (for comparison, California has about 13.8 million households). As cloud computing has soared, electricity prices across the US have (predictably) increased by 25-30% in just the last five years.
This places our elected leaders in a very difficult position. Say yes to the data center – and watch energy prices for your constituents explode or say no and lose out on an important investment opportunity.
But what if there was a third option – an elegant solution that solved a lot of different problems for both people, our planet, and the hyperscalers that needs serious consideration?
Renaissance (Resilience)
Instead of building one new giant power plant to feed a data center (15 years for a nuclear power plant or up to five years for a natural gas plant), a company called Resilience is doing something much faster and much more interesting.
They’re putting solar panels, batteries, and energy efficiency investments into ordinary American homes and businesses, thousands of them, then bundling all of that electrical capacity together into one giant “virtual power plant“ that can sell all that power back to the grid on demand. Because these homes are already wired into the grid, this power doesn’t have to wait in line behind the multi-year interconnection backlog that’s currently blocking $98 billion in stalled data center projects. It can come online in 3 to 9 months. The legal instrument the data center operator signs with Resilience is a long-term energy procurement contract called a power purchase agreement.
“No one complains about solar panels or batteries when Facebook is giving them to you for free.”
Resilience estimates there’s 1,000 gigawatts of untapped rooftop solar potential sitting on top of American homes right now, more than ten times what AI is projected to need by 2029. Only 5 percent of eligible homes have solar installed today.
But this is what makes this model so transformative. In the old model, a data center moves in, electricity prices near it jump, and the community absorbs the cost while someone else (Meta, Google, Amazon, et al.) keeps the profits.
Resilience flips that.
The homes with the “distributed energy resources” get extremely low energy bills – maybe even $0 per year. Local roofers, electricians, and developers get more work and new jobs. The power stays clean and is fast to build, and for once the people bearing the cost of America’s AI buildout are the same people enjoying the benefits of it.
Beyoncé Laws
There’s an adjacent policy mechanism moving through statehouses right now, and yes, everyone in the industry actually calls it BYONCE, short for Bring Your Own New Clean Energy. If a data center wants to plug into the grid, it has to bring new clean energy generation sufficient enough to cover its own new demand. Washington state has bills moving. Pennsylvania’s governor built it into his GRID standards. Illinois environmental groups are pushing it. And this is exactly where the VPP plus PPA model I just described shines.
A traditional utility-scale plant takes years to satisfy a BYONCE requirement. A distributed network of home solar and storage, the kind Resilience is building, can come online in months because it skips the interconnection queue. BYONCE sets the rule. The VPP plus PPA structure is how the data center developer complies with it.
Formation
It’s time for Big Tech to get information and get with the VPP plus PPA program. For years, natural gas was just cheaper than solar and batteries, full stop. That’s no longer true.
Solar plus battery storage now prices at roughly $74 per megawatt-hour; new natural gas runs about $70. Meanwhile, the cost of a gas turbine has tripled in three years, now $2.2 to $2.5 million per megawatt, and the backlog to get one built runs 3 to 6 years. Solar and storage can be up and running in 12 to 18 months. For a hyperscaler, every year a data center sits waiting on power is a year of AI revenue not captured, easily worth hundreds of millions on a large campus. Add in the political cover that comes standard with this model, showing up with lower bills and local jobs already built in instead of another gas plant proposal, and the case stops being about saving pennies on the megawatt hour. It’s about who gets to flip the switch first.
Hold Up (Save the H20)
The 97 terawatt-hours of fossil-fueled data center energy require vast amounts of water to heat up into steam (from burning gas or coal) to turn turbines and generate electricity. Much of this water is returned to lakes and rivers (albeit at a much higher temperature that isn’t good for native species). But a lot of this fresh water evaporates and can never be used again. Displace that fossil generation with solar and storage instead, and you avoid roughly 31.7 billion gallons of water consumption a year. That’s something like 48,000 Olympic swimming pools, gone from fresh water demand, every single year, just by changing how the electrons get made.
Freedom (Reducing Everyone’s Carbon Footprint)
Run the same math on greenhouse gas pollution, and the numbers are staggering. Converting that fossil-fueled share of data center power to solar plus storage avoids roughly 54.8 million metric tons of CO2 a year. That’s about 12 to 13 million gasoline-powered cars taken off the road, every year, indefinitely, just from how one industry chooses to power itself. It also happens to solve one of the hardest decarbonization problems in the book, residential and commercial buildings, which most climate plans struggle to address.
The Choice Ahead
The opportunity in front of the data center industry is not simply to find more power. It is to procure power in a way that earns public trust, accelerates deployment, and leaves host communities better off. Distributed solar, storage, efficiency upgrades, virtual power plants, and long-term clean energy procurement can turn data center growth from a source of local anxiety into a source of local benefit.
Matthew X. Roling is the Founding Executive Director at Abrams Climate Academy, Kellogg Professor & Former KPMG Climate Advisor