Imagine trying to power the entire country of Denmark – twice – just for the computers that run ChatGPT, Google Gemini, and every other AI model. That’s essentially what’s coming to the U.S. grid over the next decade, and one company just stepped up to say: “We’ve got this.”
On Monday morning, the CEO of America’s largest clean-energy generator basically told Wall Street that the AI power crisis isn’t a problem – it’s the biggest growth opportunity utilities have seen in fifty years.
The 15-Gigawatt Wake-Up Call Nobody Saw Coming
John Ketchum didn’t mince words. His company plans to originate at least 15 gigawatts of brand-new generation capacity whose sole customer will be data center hubs scattered across the United States by 2035. To put that in perspective, 15 GW is roughly the installed capacity of the entire nation of Sweden. And Ketchum immediately followed it up with the line that made analysts sit up straight:
“Quite frankly, based on what we’re seeing today, we’ll be disappointed if we don’t do more.”
– CEO John Ketchum, December 2025 Investor Day
He then casually mentioned there’s realistic upside to double that number – 30 GW – if current customer conversations keep trending the way they are right now.
Let that sink in for a second. One single utility-regulated developer is talking about adding the equivalent of thirty large nuclear reactors’ worth of power in ten years, almost entirely for hyperscale computing campuses.
Why This Matters More Than Any Other Energy Story Right Now
We’ve all read the headlines about AI’s insane electricity appetite. Reports from grid operators warning about blackouts. Investment banks throwing around estimates that data centers could eat 8-12% of U.S. power by 2030. But most of those pieces end with a worried shrug.
What happened Monday was different. It was the first time a major player stood up and said, “We already have the sites, the permits, the capital plan, and the customers lined up. This is happening.”
In my view, this single presentation might be looked back on as the moment the market finally priced in that the AI build-out isn’t going to be constrained by electricity supply – it’s going to drive electricity supply.
The Google Partnership: First of Many?
Alongside the 15 GW announcement came something equally eye-opening: a joint development agreement with Alphabet’s Google unit to build multiple gigawatt-scale data center campuses across the U.S., with optionality to expand further.
These aren’t your grandfather’s server warehouses. We’re talking about campuses that individually pull 1–3 GW of continuous power – the kind of load that used to be the province of entire metropolitan areas.
And here’s the part that quietly blew my mind: the power mix for these sites will be everything. Solar, wind, battery storage, new natural gas, and – in at least one case already public – restarted nuclear capacity.
Remember the Duane Arnold nuclear plant in Iowa that shut down in 2020? It’s coming back online specifically to feed Google data centers under a long-term power purchase agreement signed in October. That deal suddenly looks like the template for a dozen more.
All-of-the-Above Energy Strategy – Finally
For years we’ve had this exhausting ideological fight: renewables versus fossil fuels versus nuclear. Meanwhile, data center operators have been remarkably agnostic. They just want terawatt-hours of firm, affordable, and increasingly carbon-free electricity, 24/7/365.
- They’ll take solar + batteries for daytime peaks
- Wind when it blows
- New high-efficiency natural gas plants for when renewables lag
- Nuclear for always-on baseload
- And they’ll pay premium prices if you can deliver it on their timeline
NextEra’s pipeline reflects exactly that pragmatism. The company already has roughly 20 GW of gas-fired projects in development – most of which will now find homes serving these new campuses. At the same time, they remain the undisputed king of utility-scale solar and onshore wind origination.
In other words, the hyperscalers are accidentally solving the energy transition’s biggest political problem: they’re willing to fund all low-carbon solutions at once.
The Affordability Question Nobody Wants to Ask
Ketchum was refreshingly blunt about the elephant in the room: regular households are already complaining about rising electric bills, and now we’re talking about adding tens of gigawatts for Big Tech?
His answer was simple and, honestly, probably correct:
“Hyperscalers can solve that problem by bringing and paying for their own power generation and infrastructure.”
Translation: if Google, Microsoft, Amazon, and Meta want another 100 GW of compute by 2035, they’re going to have to write the checks for the power plants themselves – either directly or through long-term contracts that make new projects bankable. Residential customers shouldn’t see their bills spike to fund Silicon Valley’s AI ambitions.
Whether regulators will actually hold that line remains to be seen, but the model is already working in places like Virginia and Texas.
What This Means for Investors
If you own traditional utilities, this should feel like Christmas came early. Suddenly the growth narrative isn’t “hope permitting reform happens someday” – it’s “we have signed LOIs for the next ten years of origination.”
For clean-energy pure-plays, the story gets even better. The same hyperscalers demanding carbon-free power around the clock are effectively subsidizing massive new renewable and nuclear build-out. Every gigawatt of data center load could unlock 2–3 GW of solar, wind, and storage.
And for anyone worried about natural gas lock-in – relax. These new combined-cycle plants are being built with hydrogen-blend capability and carbon-capture readiness. They’re bridges, not destinations.
The Bigger Picture Nobody Is Talking About Yet
Here’s the thought that keeps me up at night: what happens when the rest of the world copies this model?
Europe is already years behind on data center power. The Middle East is desperate to diversify from oil. Southeast Asia wants to become the next AI hub. Every one of those regions is watching how the U.S. solves this problem.
If American developers who crack the code – site selection, co-location, mixed-fuel campuses, 36-month delivery timelines – are going to export that playbook globally. And the company that already has 15–30 GW of domestic projects lined up? They’ll probably be leading that charge.
We’re not just witnessing the build-out of AI infrastructure. We’re watching the birth of a new kind of global energy developer – one that pairs Big Tech balance sheets with utility execution at nation-state scale.
And it starts with 15 gigawatts that might quietly become 30.
The grid of 2035 isn’t going to look anything like the grid of 2025. And after Monday, I’m a lot more confident we’ll actually get there on time.
