Holtec IPO

Holtec is the first nuclear IPO I have seen with an actual operating business behind it. I still think I'm going to pass on this IPO, though.

In this post, I'll go through why I think I'm going to skip it.

Here's a link to their S1: https://www.sec.gov/Archives/edgar/data/2104277/000119312526301023/d40440ds1.htm

First, who tf is Holtec?

Founded in 1986, by Dr. Krishna Singh (who is still founder, CEO and chairman), Holtec partners with 150 commercial reactors as clients.

They have 3 manufacutring plants in NJ, PA, and OH with over a million square feet.

Their current business model has made them known as the "undertaker of nuclear power", and they have the first NRC-approved SMR reactor. Both of these in a moment.

The terms of the IPO: 50 million shares, priced between $15 and $18, with a target amount raise of $900 million, implying up to a $10.2 billion valuation

There's some governance things while reading the S-1, I'll cover that below as well.

How they make money today - The Undertakers of Nuclear Power

When a nuclear power plant has spent fuel, they take the fuel rods out of the reactor and store them in a pool, also known as wet storage. After it has cooled down (radioactively), it gets put into dry storage.

Holtec Wet and Dry Storage

Holtec has this market all but cornered. They claim more than 90% of U.S. wet storage and about 75% of U.S. dry storage for the entire market.

According to the company, they believe they have an extremely sticky product. They claim that no customers ever switched away from its dry storage. And as more plants are built, and nuclear power continues to grow, they will enjoy growth along with that trend, assuming the US does become more nuclear friendly, which it is seeming that we are.

The second way they make money today is just as fascinating. Since 2018, the company has done a "acquire the plant and decomission" strategy.

Think of it as a prepaid funeral.

Every nuclear plant in America has a decommissioning trust. For decades, a few cents on every customer's electric bill went into that fund so there'd be money to tear the plant down when it died. Pilgrim's trust was filled by Boston Edison customers over the 28 years the utility owned the plant.

Holtec's move: when a plant shuts down, buy it. The plant comes with the trust fund and with the legal obligation to do the cleanup. Do the cleanup for less than what's in the fund, and the leftover is yours. The S-1 says it plainly: revenue from decommissioning has historically come from management and drawdown of the decommissioning funds. They're the largest owner of shuttered reactor units in the United States.

There's a second check in the mail, and it comes from Uncle Sam. Under the Nuclear Waste Policy Act, the federal government was supposed to start hauling away spent fuel in 1998. It never built the repository. So the fuel sits in casks at the plant, the owners sue, and the government pays their storage costs. That's why Holtec expects DOE reimbursements for babysitting fuel the government promised to take. America's failure to build a dump is a Holtec profit center.

The catch runs both ways. If the cleanup costs more than the fund has, the owner eats the difference. Hold that thought, because it explains a move the founder made right before this IPO.

An example of this is the Pilgrim facility. Holtec took Pilgrim in 2019 with more than $1 billion sitting in its trust. Their own filings projected about $252 million left over when the job was done, plus DOE reimbursements for spent fuel that could add several hundred million more. The latest numbers Holtec filed with Massachusetts: $484.7 million in the trust at the end of 2024, about $557 million of remaining cost until the last fuel leaves the site in 2063, roughly $247 million projected in the fund when major decommissioning wraps in 2035, and about $75 million left at the very end, assuming a 2% real return the whole way. Same slides spell out the trigger points where Holtec has to make up any shortfall.

A prepaid funeral, an undertaker who works cheap, and a government that pays you for storing the body. That's the business that made Holtec's money for the last seven years.

Holtec's Pilgrim plant they decomissioned for profit

source

Their SMR reactor could be huge

SMR stands for small modular reactor. Palisades is 805 megawatts. Holtec's SMR-300 is designed to put out about 300 to 340 megawatts per unit, so roughly a third the size, built in factory modules instead of poured on site, and planted in pairs. Two of them, Pioneer One and Pioneer Two, are slated for the Palisades site.

What's actually new about it, and what isn't. The physics isn't new. It's a pressurized water reactor, the same basic design that's run the US fleet and the Navy's ships for 70 years.

The company calls it Gen III+, which means evolutionary improvements on the design we already know how to license, which Holtec sees as a feature, not a bug.

The new part is the safety architecture and the footprint. Holtec calls it "walk-away-safe": after a shutdown or an accident, the plant can sit in safe shutdown indefinitely with no operator action, using passive systems and the water already inside it to carry off decay heat. Decay heat is the heat a reactor keeps making after you shut it down; getting rid of it is the whole game. Credit to Holtec for adding, in their own filing, that "walk-away-safe" is not a regulatory term.

The footprint is the sales pitch. About 15 acres for a single unit, and by their design math two dual-unit plants, four reactors total, could replace a 1,200 MW plant with over 1,300 MW, with a roughly three-year build target.

How big could it be for the business? Today, zero. The S-1 says the SMR-300 is at a pre-construction stage: technical design done, site picked, financing partial, dirt moved. Nobody has built one. The plan: Pioneer One and Two at Palisades with Hyundai E&C and Mitsubishi Electric as partners, a $400 million DOE grant that still depends on a funding agreement getting signed, then Oyster Creek in New Jersey, where the license termination plan was approved July 30 and the plan calls for four SMR-300 units, then a non-binding deal with Entergy to look at projects across the Gulf South. Singh told the FT the first two should be producing power by 2031... yikes

They also have a "coal-to-clean" pitch: take a retired coal site, drop in SMR-300s plus their solar and thermal storage tech. Worth knowing that the storage and solar products are at technical design, not deployed, and licensed to Holtec Asia, a company owned by the founder personally. More on him later.

Here's the number that tells you how they'll pay for all this. They can't. Not alone. The S-1 says their development arm plans to raise money by selling equity in projects at de-risking milestones, and that they're likely to sell some of their stake in the Palisades SMRs. They will sell pieces of the reactors to build the reactors. Which is fine. It's also why you're being offered shares.

The Megatrend of Holtec

Every one of these numbers comes from Holtec's own S-1, which is exactly where a company puts the numbers that make the roadshow easy.

Global electricity demand up 75% by 2050 from 2024 levels, per BloombergNEF. Global data center demand 5.5 times higher by 2040. In the US, the annual additions to the grid from data centers and industry roughly 15 times higher by 2040 than in 2025. Note that last one carefully: it's the yearly additions that go 15x, not total demand. The administration's stated goal is to take US nuclear from about 100 gigawatts today to 400 by 2050. Worldwide, 417 reactors are running and 77 are under construction. The US runs 94 reactors for about 97 gigawatts, and 73% of US sites are planning or considering power uprates, more than 5 gigawatts in total.

I believe the trend. AI needs baseload power, baseload power means gas or nuclear, and the politics of nuclear have flipped faster than I've seen anything flip in energy. The problem with a trend everyone believes is that it's already in the price. Every nuclear name on the tape is priced like 2040 arrived last quarter. More on that in the numbers.

They may be the first to restart a closed nuclear plant

Palisades plant, which Holtec intends to restart

Palisades went online in 1971 on the shore of Lake Michigan. Entergy announced in 2016 it would close it, and shut it down on May 20, 2022, after 50 years, even though the plant was licensed to run until 2031. Holtec bought it to bury it. Then, in the company's own words, a dramatic change in the public's and governments' posture toward nuclear happened, and they decided to resurrect it instead. Nobody in this country has ever restarted a commercial reactor after a permanent shutdown. If they pull it off, it's a real first.

The money behind it: a Department of Energy loan guarantee facility of up to $1.52 billion, a $300 million grant from Michigan, Holtec's own capital, and 28-year power purchase agreements with regional electric co-ops that lock in the revenue.

Now the part I care about, because I spent years running a reactor plant in the Navy. On August 30, Holtec started loading fuel, which put Palisades in Mode 6. Commercial plants live in six modes. Mode 6 is refueling: fuel in the vessel, reactor shut down. Then cold shutdown, hot shutdown, hot standby, startup, and finally Mode 1, power operation. Fuel loaded is a milestone. It is the first of six.

The schedule has slipped every time someone put a date on it. When the DOE loan closed in 2024, the expectation was a late 2025 restart. Steam generator problems pushed that to mid-2026. Then the language softened to "when the plant is ready for long-term operations." Singh told the Financial Times he still expects it back this year. The hard stop is the contract: power to the co-ops by March 2027. Steam generators are the classic old-plant problem, thousands of thin tubes that crack and corrode, and fixing them is slow, expensive work. I'm not saying they won't make it. I'm saying they own the outcome to their creditors and their customers, the clock is real, and "fuel loaded" is not "power on the grid."

the Numbers.

This is where I stopped being excited.

Revenue fell from $765.5 million in 2024 to $576.6 million in 2025. Down 25%. Income from operations fell from $291.0 million to $37.1 million. Down 87%. In the twelve months before they decided to sell you shares.

Part of what made 2024 look so good wasn't cash at all. Inside operating income sits something called an asset retirement obligation gain: $177.7 million in 2024 and $58.2 million in 2025. Plain English: the ARO is the company's estimate of what the cleanup will cost. Lower the estimate, and the difference shows up as profit. No money changed hands.

Holtec's own explanation, through the Philadelphia Inquirer: 2024 was unusually profitable thanks to extra revenue from shutting down Indian Point and the timing of payments, and the company expects US decommissioning work to decline in the near term because fewer plants are retiring. Read that again. The undertaker business shrinks when fewer plants die, and fewer plants are dying because the trend they're selling you is real.

2026 so far: first-quarter revenue of $165.3 million and net income of $17.8 million, versus $177.7 million and $25.4 million a year earlier; first-half revenue of $269.9 million, down from $286.6 million. Trailing twelve-month revenue is about $560 million.

So what are you paying? At the top of the range, $10.2 billion for $560 million of revenue, about 18 times sales, for a business that's currently shrinking, plus a reactor that doesn't exist yet. For comparison, Oklo trades near a $15 billion market cap with no revenue, and X-energy came public in April at $23, popped to $29.20, and hit a fully diluted value just shy of $12 billion, also with no revenue. Holtec is either a $560 million industrial business with a free reactor lottery ticket, or a reactor startup priced like the others with a cask business bolted on.

One more number for the people who buy on day one. Oklo hit an all-time high of $193 in October 2025. $10,000 invested at that high is worth about $2,300 today.

Founders and governance

Two rules I keep in my head for IPOs. Munger: show me the incentive and I'll show you the outcome. Buffett: in an IPO the seller picks the moment and knows more than you do, so bargains are rare by design. Now look at what Dr. Singh built.

Votes. Class B shares carry 10 votes each. The sunset on that power is not triggered by the founder's death or by transfers within the family. Holtec will be a "controlled company" under Nasdaq rules. You buy the car; he keeps the keys; his kids inherit the keys.

The structure. This is an Up-C, a two-story house where the public owns the upstairs company and the actual business sits downstairs, still partly owned by the people selling you the upstairs. It comes with a Tax Receivable Agreement: the public company will pay 85% of the net cash tax savings it realizes to the founder's entities. He sells you the house and keeps most of the tax deduction.

The cash out. Part of the IPO proceeds gets distributed to Holtec Holdings so the founder's trusts can pull out the previously taxed net income of the old S-corp on a tax-free basis. That's a number I want to see before I decide anything, and it's a number that wasn't filled in the version of the filing most people read.

The disclosure. Holtec files as an emerging growth company: only two years of audited financials and no auditor attestation on internal controls. For a 40-year-old company.

The family. The founder-controlled affiliate Holtec Asia supplies condensers and holds the licenses to the storage and solar tech. In 2025 Singh took $7.5 million in total pay, his wife Martha, the vice chairman and chief strategy officer, $856,000, and their daughter, the communications director, $871,000. The Camden headquarters is named the Krishna P. Singh Technology Campus.

The history. In 2010 the Tennessee Valley Authority barred Holtec from federal business for 60 days and fined it $2 million after an investigation found about $54,000 funneled to a TVA employee's shell company. In 2014, applying for a $260 million New Jersey tax break, Singh answered "no" to whether the company had ever been barred, which triggered a state investigation and years of litigation before New Jersey's Supreme Court declined in 2024 to disturb Holtec's win. Holtec called it an inadvertent omission and the courts sided with the company. Both of those things are true. You decide how much weight to give a sworn "no."

And the funerals. Remember the prepaid-funeral business? Right before this IPO, Holtec moved it out. Under something called the NAMCO restructuring, the shuttered plants they own (Pilgrim, Indian Point, Oyster Creek, Big Rock Point), along with their trust funds and their cleanup liabilities, are being transferred to DEAMCO, a subsidiary of Holtec Holdings, which is the founder's company, not the one you'd be buying. Palisades stays. The stated reason is to reduce exposure to market swings in the trust funds and to cost uncertainty. Fair enough. The effect is also that the sites that produced the 2024 windfall leave the public company before you own it, and the public company keeps only the service fees for doing the work. The S-1 has pro forma financials that show the company after this move. That's the company you'd actually own.

The re-org makes things really complicated too... here's an image from the S-1 for how the company was structured before and after.

Yikes. That's confusing. Count me out.

Conclusion

I'm not passing because nuclear won't work. I think it will. I'm passing because I can't answer four questions yet: what the real net income was, how much of my money goes to the founder on day one, what the business looks like after the funerals leave, and when Palisades actually makes power. Buffett's rule is that you don't have to swing.

What do y'all think?