Going Nuclear: AI's New Power Partner
The risk and reward of the MSFT, AMZN and GOOGL AI energy pacts ...
It was early 1989, I was working as a business reporter for a newspaper in York, Pa., and I found myself up at the Three Mile Island (TMI) nuclear plant near Harrisburg.
It was the 10th anniversary of the accident named for that power plant. I was working on a package of stories — about the accident, the still-underway cleanup and the obvious “what-comes-next” analysis of America’s future. A public-relations “flak” for plant owner General Public Utilities took me and a York Daily Record photographer on a tour of the still-operating “Unit 1” and gave us the most-current lowdown on the failed “Unit 2.”
To this day, I still remember watching on a black-and-white TV screen as a robot worked inside the damaged reactor core of TMI-2; even a decade after the accident, Unit 2 was still far too “hot” for human workers to brave.
Nuclear power is back in the news …
Even Three Mile Island …
For a guy like me, it’s kinda like I bolted a “Flux Capacitor” in my old Hot Rod (or stepped into Dr. Who’s TARDIS phone booth) and jumped decades back in time.
Maybe all the way back to the 1970s.
You see, nuclear power is a thread that’s run through a big swath of my life.
My Dad, William Patalon Jr., was a career engineer — with two degrees in physics. Most of his career was spent in the defense/aerospace business — the early part with United Technologies’ Hamilton Standard and Pratt & Whitney units, and the long last swath was with the electronic systems unit of Westinghouse Electric Corp., which was later sold to Northrop Grumman Corp. NOC 0.00%↑.
In the middle of that, my Dad spent about eight years with Westinghouse’s commercial nuclear business near Pittsburgh.
In short, first thanks to my Dad — and then later due to my four decades as a reporter and financial writer/analyst — I’ve had a ringside seat to both the New Cold War … and the global nuclear-power saga.
And nuclear power was exactly that … a saga.
But with nuclear energy gaining a new currency here in the AI Era, my story is one you need to know. To power all the high-powered data centers we’ve been telling you about, we’re going to need some high-powered energy sources.
In the last few weeks alone, Microsoft Corp. MSFT 0.00%↑, Amazon.com Inc. AMZN 0.00%↑ and Alphabet Inc. GOOGL 0.00%↑ have announced AI-related nuclear deals — including one (believe it or not) involving Three Mile Island.
But it won’t be the immediate slam-dunk most folks are expecting …
Believe me … I know.
As well as anyone telling you the “AI Nukes” story …
LET’S MAKE A DEAL …
Back in September, Constellation Energy Corp. CEG 0.00%↑ announced plans to restart Unit 1 (which was shut down in 2019) — after Microsoft signed a 20-year deal to buy power from it.
Earlier in October, Google inked an AI-related pact to buy nuclear-related electricity from the privately held Kairos Power.
And Amazon said on Wednesday it has signed three agreements to develop nuclear power projects.
And trust me, the need is there.
Thanks to AI, digitalization and massive data-storage needs, data center power demand will nearly triple (surge 160%) by 2030. Those massive processing operations currently consume 1% to 2% of all power — but could account for 3% to 4% of electricity production by the end of the decade, says Goldman Sachs Research. S&P Global says that share could rise as high as 9%. And consultant McKinsey & Co. says it could hit 11% to 12% of power demand.
Folks who obsess over the recent AI demand headlines are missing another element of the American energy story — an element that’s not as sexy, but might actually carry far more urgency.
I’m talking about the aging and stressed U.S. power grid — the plants, transformers, and power lines that are too old, brittle and unreliable to handle the growing power demand … or the extreme weather that we’re expected to keep seeing. Blackouts, brownouts, power surges and other issues are red-flag warnings of what’s to come – especially as they increase in frequency and duration.
Addressing these issues and modernizing “The Grid” — upgrading the infrastructure, bringing renewables online and adapting “smart-grid” technologies — will cost an estimated $1 trillion to $2 trillion over the next decade alone.
And nuclear power is part of that upgrade.
Last year, nuclear energy accounted for about 18.6% of America’s power generation. Though green extremists will likely disagree, that represents 48% of the country’s clean electricity.
Growing up as the son of an engineer who was a player in the nuclear business, I saw – and still see – the benefits of atomic power. It’s not “if.” It’s “when.”
But history tells us that “when” may be further away than folks are thinking …
“NO NUKES, NO NUKES!”
Coming out of the 1960s, nuclear offered the promise of “clean” energy. That “promise” was supercharged by an “urgency” created by America’s dependence on foreign oil, growing fears of air pollution, an inflationary cycle that had more lives than Michael Myers and the promise of continued economic growth. (Real gross domestic product (GDP) averaged 5% in the ‘60s, payrolls increased a postwar high of 32%, and government tax revenue surged 65% from 1965 to 1970).
Two major 1970s energy crises — the first in 1973 and the second in 1979 — added to nuclear’s allure as a “cheap, clean and reliable” energy source. Indeed, most of the commercial reactors that have operated here in America were under construction by 1974.
And the 1970s is where my story resumes.
As energy worries revved up commercial interests in nuclear power, the anti-nuke movement got busy, too. The Friends of the Earth, the Sierra Club and the Ralph Nader-inspired “Nader’s Raiders” disputed the safety of the technology, and worked to slow or stop them altogether.
Hollywood did its part in the protests with flicks like The China Syndrome (1979) and Silkwood (1983). (In a weird twist, The China Syndrome debuted just two weeks before the Three Mile Island accident.)
In very late 1978 or early 1979, as a senior at Maryland’s Bel Air High School and co-editor of The Bellarion school paper, I wrote an editorial about the need for commercial nuclear power.
Talk about lousy timing …
On March 28, 1979, Three Mile Island suffered a partial meltdown.
Back then, there were really only three TV stations (ABC, NBC and CBS … CNN was launched in June 1980) in any community. And it was: “All Three Mile Island, all the time.”
My Dad had long ago moved back into the defense business. But we lived just 63 miles from the plant — so naturally my folks worried about the fallout if a total meltdown played out.
That didn’t happen. But my Dad and I both knew that, for America’s nuclear energy sector, Three Mile Island was the day the music died.
And a decade later, during my 1989 reporting tour, as I watched that flickering TV screen, I knew that nuclear energy’s potential was as ruined as the TMI-2 core that robot was scooping away at.
A STILLBORN SECOND CHANCE
In music, artists like Madonna, Cher and Prince have earned “single-name status” — in a good way.
Nuclear power is the single-name antithesis — thanks to TMI, Chernobyl (1986) and Fukushima (2011).
Atomic power had a second chance there in the 2000s — experts called it a “nuclear renaissance” — because of smaller, compact reactor designs.
In 2002, in fact, U.S. President George W. Bush launched the “Nuclear Power 2010 Program,” hoping to jump-start reactor construction and to bring plans online at decade’s end. That didn’t happen, and any hoped-for rebirth ended on March 11, 2011, when a tsunami wrecked the nuclear plant in Okuma, Japan.
As the disaster unfolded, my Dad and I were in Ohio, on one of the trips we took each year; I remember him spending an hour or so every morning on a call as part of a problem-solving working group he’d been invited into – not a surprise, given his ability to analyze and troubleshoot technical problems in a way that helped folks “get to the essence.”
Six years later, in March 2017 (maybe there’s a message here about avoiding nuclear projects in March?), Westinghouse — my Dad’s old company and the No. 1 nuclear plant builder — filed for bankruptcy, after losing $9 billion from two projects.
I’ve shared these stories — my personal journey — for a reason.
Like that Hollies’ song says, the nuclear power industry’s road “is long … with many a winding turn.” But there are opportunities. Big opportunities. And the very definite “takeaways” I’ll bring you can help you navigate the Next Nuclear Revolution.
FIVE TAKEAWAYS … AND “WHAT’S NEXT
Here we are … again on the doorstep of a nuclear-power renaissance … the “third chance” in America (if you include the 1970s).
If you’re an investor watching the excited response to all the recent nuclear deals, here’s what you need to know before investing.
No. 1: Nuclear power is still the answer: Fusion is still decades away. Wind and solar are intermittent power sources. But nuclear is a “baseline” power source. It’s clean. It’s cost-effective. And once you get it up and running, it’s efficient. In that sense, it’s a great “partner” with natural gas. And, given the demands we see, the aging power grid, threats from extreme weather and the desire to move away from fossil fuels, it’s the single-best option we have. That’s how I felt when I wrote that school editorial in 1978. And it’s how I feel today.
No. 2: But it’s not necessarily the “straight line” the euphoric headlines (and stock prices) are signaling: Google said its first Kairos reactor would be up and running by 2030 — with more reactors going “live” five years later.
Call me a Contrarian (which I am). But I think that’s a hyper-optimistic timeline. Granted, these modular reactors are a newer, streamlined technology. But new also means “need time to prove.” And it’s not just construction. It’s the regulatory approvals, too.
Want an example?
Southern Co. SO 0.00%↑ is a case study for new power plant construction. Southern’s Vogtle Units 3 and 4 benefited from Units 1 and 2. The expansion was initially proposed in early 2004. The Nuclear Regulatory Commission (NRC) approved the construction and operating licenses on Feb. 9, 2012, and the target for completion was 2020. But Unit 3 didn’t begin commercial operations until July 31, 2023, and Unit 4 didn’t start until April 29 of this year. Even worse: The project’s initial projected cost was $14 billion; its actual cost was $30 billion. So it took two decades and double the expected cost from start to finish.
Against that backdrop, Microsoft was shrewd in striking the TMI deal. That pact sidesteps the risks, delays and costs of new construction. Even so, Constellation says it needs to spend about $1.6 billion to revive the plant – and expects it to be operational by 2028. As of his moment, however, none of the federal, state or local regulatory permit applications have been filed, reports state.
No. 3: In algebra, they’d call this “Protests Cubed:” I told you about the environmental protests that spooled up in the 1970s. Back here in the “present” timeline (I hit 88 mph and made it back), we’re already seeing competitive protests. And AI protests. That’s what I mean by “Protests Cubed” — or protests to the third power.
In terms of competitive protests, a pact that Amazon stuck earlier this year with Talen Energy TLN 0.00%↑ has been challenged by a coterie of rivaling utilities who contend the deal will stress the grid and spike consumer-electricity costs.
As nuclear projects get going, expect more consumer protests — perhaps even lawsuits — which (just like the old days) could slow the permitting process.
Just look at the photo I took. The group is protesting a major power transmission line project that’ll cut across a big swath of picturesque land in Maryland. The project’s purpose: To deliver electricity to a power-hungry data center development. (The protest placard, in an equestrian pasture, with a rural graveyard backdrop, really says it all.)
Combine nuclear power plants and data centers … and who knows how virulent the protests will be. Or to what degree those protests, court cases or environmental-impact studies will slow the projects down.
The potential for such scuffles is high: I know from my talks with energy-sector experts that engineers and financiers want those data centers as close as possible to major “population centers.” And as I know from my years as a reporter (covering annexations and developments), people are predictable: They want technology, they want fast Internet, they want AI’s benefits — they just want the “enabling technology” (and especially nuclear reactors) “somewhere else.”
No. 4: Any problems will trigger “outsized” blowback: Following Three Mile Island, more than 100 orders for nuclear reactors — including some already under construction — were canceled in the ‘70s and ‘80s. Some companies were bankrupt as a result.
After Fukushima, there was the Westinghouse bankruptcy — followed by another 15-year “nuclear winter” of sorts.
Any major technical problem this time around could lead to that same kind of nuclear moratorium.
No. 5: The Bottom Line: The AI Era is a huge storyline: So there’s major money to be made here. But you also need to manage risk.
Investor reaction to these nuclear deals has been euphoric. But the payoff is still down the road, meaning it’s not a slam-dunk payoff.
So caution is the watchword for nuclear, especially as there are plenty of AI-related “power plays” to consider.
We recently put together a nifty report on seven stocks — a great starting point for your own spade work.
But I’ll be back with a detailed look at some specific investments soon.
See you then;
