Concentration Risk

The event nobody saw coming

In nuclear power, the events that permanently rewrote the rules were almost never the ones the industry was defending against. Markets work the same way. For a utility career built on a single stock, that is not a metaphor — it is the whole risk.

Two Minute Portfolio Preparedness Review Contact

The Pattern

It is always the rule nobody thought to write

The American nuclear rulebook was not written in advance. It was written afterward, in the wreckage, one surprise at a time — and never once about the accident the industry had been drilling for.

Markets are assembled the same way. Every rule governing trading, disclosure, and risk today exists because something happened that the people running the system considered impossible the week before it happened.

Five of them. The stories are worth telling properly, because the shape of each one should be familiar to anyone who has watched a portfolio.

01

The Candle

Below the control room at Browns Ferry there is a space called the cable spreading room, where thousands of electrical cables fan out from the panels above and run through the wall into the reactor building. The gaps around them were packed with polyurethane foam.

Two men were checking those seals for air leaks. The accepted method — the method everyone used, the method nobody questioned — was to hold up a lit candle and watch which way the flame bent.

The flame found the foam. They tried to beat it out and pushed it deeper into the wall instead. The fire got into the cable trays and traveled through into the reactor building, and it burned for roughly seven hours.

Somewhere around 1,600 cables were destroyed. Six hundred of them were safety cables. In the control room, indicator lights went dark and dead in banks, and the operators of Unit 1 discovered they had lost the emergency core cooling system entirely — the pumps, the controls, the ability to know what was happening. They kept water on the core by improvising with pumps that were never intended for the job.

Browns Ferry had defenses against every accident anyone had modeled. It had no rule about candles. Five years later the NRC published Appendix R, and fire protection became a nuclear engineering discipline, because two men had gone looking for a draft.

The same failure, in markets

October 19, 1987. The Dow lost 22.6% in a single session with no news to explain it. The accelerant was portfolio insurance — a product sold to institutions as downside protection, which responded to falling prices by selling automatically, which pushed prices lower, which triggered more selling. The safeguard was the fire. Circuit breakers exist today for the same reason Appendix R does.

The thing installed to protect you is often the thing that propagates the failure.

02

The Light That Lied

At four in the morning, a blockage in a water-cleaning system shut down the feedwater pumps at Three Mile Island Unit 2. The turbine tripped. The reactor scrammed. All of that was normal. All of that was designed for.

Pressure spiked, and a relief valve on top of the pressurizer opened to vent it, exactly as intended. Then the valve stuck open. Reactor coolant began pouring out of the top of the system, and it kept pouring for more than two hours.

In the control room there was a light for that valve. The light was wired to the signal telling the valve to close — not to the valve. The signal had been sent. The light was off. As far as every man in that room could tell, the valve was shut.

What they saw instead was rising water level in the pressurizer, and they had been trained their entire careers never to let the pressurizer fill solid. So they did the responsible thing: they throttled back the emergency cooling water that the plant was automatically injecting into the core.

They starved the reactor to satisfy an instrument. About half the core melted. The containment held, and almost nothing got out, and it did not matter at all — the industry's future was decided anyway. Reactor orders stopped. And twelve days earlier, a film called The China Syndrome had opened in theaters, in which a fictional meltdown renders an area "the size of Pennsylvania" uninhabitable.

The same failure, in markets

2008. The tranches were rated AAA. The risk models were green. What the rating actually reported was not the condition of the loans — it was the output of a model that had been told housing could not fall everywhere at once. Investors, like the operators at TMI, took action based on a readout that described the instruction rather than the reality. The light was off. The valve was open.

Your dashboard reports what it was built to report. It does not report what is true.

03

The Rust

For years, crews at Davis-Besse had been cleaning a reddish-brown crust off the top of the reactor vessel. It was on the head, it was around the control rod nozzles, and there was a lot of it. They scraped it, they logged it, they called it rust, and they went home.

It was boric acid. It had been weeping out through cracks in the nozzles, and it had been eating the reactor vessel head — six and a half inches of carbon steel — for something on the order of six years.

In March 2002, an inspector put a tool into what he thought was a shallow pit, and it kept going. When they opened it up they found a cavity roughly the size of a football, corroded clean through the steel. What was left holding back the primary coolant system, at over two thousand pounds per square inch, was a stainless steel liner about three-sixteenths of an inch thick. It was already bulging outward.

The plant had been passing its inspections. Every one of them. The rust was not hidden. It was on the head, in plain view, for years, and it was catalogued and cleaned and dismissed, because nobody asked the one question that mattered: where is this coming from?

The NRC later called it the most serious safety issue in the United States since Three Mile Island.

The same failure, in markets

Silicon Valley Bank, March 2023. The losses that killed it were not hidden in a derivative or buried offshore. They were unrealized losses on Treasury bonds — the safest instrument that exists — and they were disclosed, quarter after quarter, in a public filing anyone could download. Everybody saw the rust. Everybody called it rust. Nobody asked where it was coming from.

The warning is almost never missing. It is almost always dismissed.

04

The Repair

San Onofre spent roughly $700 million on new steam generators. The point was to extend the plant's life — twenty more years of clean baseload for Southern California. The new generators were bigger, with more tubes, and the design changes altered how steam and water moved inside them.

Within a year the tubes were rubbing against each other and wearing through. On January 31, 2012, one of them ruptured. When they went in and looked, they found wear across thousands of tubes in machines meant to last decades. The plant never generated another watt. In 2013 it was retired permanently — 2,200 megawatts gone, killed by the thing that was supposed to save it.

Crystal River in Florida died the same death, from the opposite direction. To swap its own steam generators, the utility decided to cut the opening in the containment building itself rather than hire the contractor who had done it elsewhere. It would save money. They cut into forty-two inches of prestressed concrete, and the wall delaminated — the layers separated. The repair induced new cracks in other sections. The estimates climbed into the billions with no guarantee it would ever work again.

The plant had sixteen years left on its license. It was retired without ever restarting.

The same failure, in markets

2022. Bonds were the ballast — the thing you add to a portfolio precisely so it will not sink. Then long Treasuries suffered a historic loss at the exact moment equities fell, and the 60/40 took a beating it was structurally designed never to take. Every retiree who "de-risked" into long duration on the way out the door discovered that the stabilizer was the hole in the hull.

The upgrade you make for safety can be the one that takes the asset.

05

The Spreadsheet

Now the important one, because it is the only one on this page that was not an accident.

Kewaunee ran perfectly and had a license good until 2033. It closed in 2013 because Dominion could not find a buyer. Vermont Yankee closed in 2014. Fort Calhoun in 2016. Oyster Creek in 2018. Pilgrim in 2019. Indian Point, which powered New York City, in 2021. Palisades in 2022.

And in September 2019, Three Mile Island Unit 1 — the good one, the twin that never had a bad day, that ran safely for forty-five years — shut down because it could not clear a capacity auction.

Not one of these plants melted down. Not one of them failed an inspection. They were safe, paid off, and carbon-free, and they were scrapped because natural gas got cheap and a spreadsheet said so.

The industry spent forty years hardening itself against a meltdown, and lost half its fleet to a price on a screen in Louisiana.

And then the price moved again. Palisades is being brought back — the first restart in American history of a plant already in decommissioning. Three Mile Island Unit 1 is being relicensed to feed data centers. The plant that was worthless in 2019 is worth restarting in 2026, and nobody saw that coming either.

The same failure, in markets

Enron, 2001. Not a crash — a single-name accounting failure at an energy company, where employees held company stock inside the retirement plan and lost the paycheck and the savings on the same afternoon. It was not systemic. It did not show up in the index. And for those employees it was the only event that ever mattered.

The thing that ends your retirement will not be the thing you drilled for. It will be a filing, a price, or a decision made in a room you have never been in.

By the numbers

$9B

Spent at V.C. Summer before the project was abandoned in July 2017. Nothing was built.

~$35B

Vogtle 3 and 4 as delivered, against an original estimate near $14 billion.

22.6%

The Dow's loss in a single session on October 19, 1987 — with no news catalyst.

42%

Year-over-year growth in data center power use across Southern's utilities, Q1 2026.

The Fork

Two utilities. One reactor. One bankruptcy filing.

On March 29, 2017, Westinghouse — the company that invented the American pressurized water reactor — filed for Chapter 11, sunk by fixed-price contracts on two AP1000 projects. Nobody in the utility sector had that in a model.

Four months later, one of those projects was abandoned. The other was finished. Same reactor design. Same vendor. Same regulatory era. Same decade.

The road not taken

V.C. Summer

SCANA & Santee Cooper — South Carolina

  • Units 2 and 3 abandoned in July 2017 after roughly $9 billion had been spent.
  • SCANA was absorbed by Dominion at a fraction of its former value.
  • The chief executive later pled guilty to fraud.
  • SCANA employees held SCANA stock.

The road taken

Plant Vogtle

Southern Company & Georgia Power — Georgia

  • Units 3 and 4 completed in 2023 and 2024, years late and billions over budget.
  • Southern absorbed the charges and a long stretch of dead money in the stock.
  • The result: roughly 2.2 GW of round-the-clock carbon-free baseload — the only new nuclear built in America in decades.
  • It arrived exactly as data centers began paying premiums for precisely that.

The most criticized capital allocation decision of the 2010s became the strategic asset of the 2020s. Nobody forecast that either.

A Southern employee and a SCANA employee, in 2015, made the identical decision. One of them is fine today. One of them is not. Neither earned the outcome.

The Committee

Who is actually managing your retirement

If a large part of your net worth sits in one company's stock, then your retirement is being run by a committee. Here is the roster, what each member decides, and how much say you have.

First, the honest part — July 2026

Right now, this committee is voting in your favor.

Vogtle 3 and 4 are running. Data center power consumption across Southern's utilities is up roughly 42% year over year. There are gigawatts of large-load projects under contract with a far larger pipeline behind them, and the capital plan runs to tens of billions through the end of the decade. The most criticized construction project of the 2010s turned out to be the strategic asset of the 2020s. The stock has been re-rated. The dividend went up again.

So nothing on this page feels urgent to you today. Your position is doing well. The board looks smart. The commission is cooperative. The demand is real.

That is not a reason to relax. That is the exact condition under which nobody acts — and it is the condition every man in every story on this page was standing in, right up until the meeting he was not invited to.

The decision
Made by
Your say
Whether the dividend is raised, held, or cut
The board of directors, in a meeting you will not attend
None
Whether the company is allowed to earn a return on tens of billions in new capital spending
The Public Service Commission, in the next rate case
One ballot, statewide, every six years
New in 2026Whether the data center demand actually shows up — and stays
Capital allocation committees at a handful of technology companies, none of which are in Georgia
None
Whether that spending comes in on budget
Contractors, a supply chain, a turbine order book, and a labor market
None
How the build is financed — and how much of the company you still own afterward
The treasury department and the equity market. Every new share issued is a share of your company you no longer own
None
What your pension lump sum is worth on the day you take it
Interest rates, set well above your pay grade and mine
None
Whether your first five years of retirement are a bull market or a bear
Nothing. No one. The calendar you happened to be born on
None

These are your portfolio managers. You have never met them. They have never heard of you. Not one of them has ever asked how your retirement is going, and not one of them ever will.

And notice what joined the roster while you were not looking. You did not spend thirty years at a utility in order to own an artificial intelligence company. But a meaningful share of today's optimism rests on the capital budgets of a few firms a long way from here. How much? You don't know. I don't know. Neither, in any honest moment, does the board.

You have exactly one decision on this committee: how much of your life savings you hand over to it.

The Objections

The two reasons you haven't done anything about this

Neither one is stupid. Both are real. And you have probably never said either of them out loud, which is exactly why neither has ever been answered.

01

"I need the dividend. I can't sell it."

This is the stronger of the two, and it deserves a straight answer rather than a lecture.

A dividend is not a wage. It is a distribution the board votes on, and boards vote both ways. The utility sector has a long record of steady payouts — and it also contains companies that cut, suspended, or eliminated the dividend when the balance sheet demanded it. A dividend is a decision, not a property of the stock.

More to the point: income is a function of the whole portfolio, not of one position. If a concentrated holding is generating the cash flow you live on, the question is not whether to give up income. It is whether the same income can be produced without one company's board holding the vote. Usually it can — and if it can't, you need to know that before you retire, not after.

02

"I helped build that plant."

You did. That is not nothing, and I am not going to pretend it is.

But loyalty to the company and loyalty to your family are two different obligations, and only one of them is holding the stock. Your company does not own shares of you. It will not adjust its capital plan because of what is in your 401(k). It has never once asked how your retirement is going.

There were people at V.C. Summer who felt exactly the way you feel. They had every reason to. The pride was earned, and the pride was real, and it did not protect a single share.

Selling the stock is not a betrayal of the work. The work is done. It is in the ground, it is on the grid, and nobody can take it from you. The shares are a separate question, and they should be answered separately.

What both of them actually are

The model became invisible

David Bohm was a theoretical physicist — he worked alongside Einstein at Princeton — and he spent the second half of his life on a question that had nothing to do with physics: why do intelligent people fail to update their thinking when the evidence changes?

His answer was that thought is not a neutral observer. Thought builds a model of the world, and then it quietly forgets that it built one. The model stops being a model and starts being reality.

“Thought creates our world and then says, ‘I didn’t do it.’”

David Bohm

Bohm identified three traits of thought that turn dangerous under pressure. Two of them are sitting on this page, in your own words:

  • Thought defends its conclusions. Evidence that contradicts the position gets minimized; evidence that confirms it gets amplified. “I need the dividend” is rarely a calculation. It is usually a conclusion looking for support — and it is a conclusion you have never had to defend, because nobody has ever asked you to.
  • Thought mistakes the past for the present. Confidence in a position often turns out to be a recording from an earlier experience rather than a fresh read of current conditions. “I helped build that plant” is a true statement about 2013. It is not a statement about 2026, and it was never a statement about your balance sheet.

Bohm called this the systemic fault: the very instrument you would use to check your thinking is the thinking. You bring a biased tool to evaluate whether the tool is biased, and you find nothing wrong. It is not a character flaw. It is how the system is wired — and it is exactly why the operators at Three Mile Island trusted the light instead of the valve.

So here is the question worth writing down. For the position you hold with the strongest conviction: what would have to be true today for this to no longer make sense?

If you cannot answer that easily about your employer stock, Bohm would say the model has already become invisible. That is not a reason to panic. It is a reason to have someone else look.

Read the full Bohm essay →

The Discipline

You already know the answer. You practice it at work.

Your industry invented defense in depth. Multiple independent barriers. No single failure permitted to take down the plant — precisely because you cannot forecast which failure you will get. You do not build a reactor around a prediction. You build it so that being wrong is survivable.

“Then you go home to a retirement plan with a single point of failure.”

Concentration in employer stock is not a view on Southern Company. It is a bet that you will land on the Vogtle side of the fork rather than the V.C. Summer side. That bet carries no edge and no information. It carries familiarity — which is not the same thing, and never has been.

And the regime is still moving. In both directions. Palisades is attempting the first restart in American history of a plant already in decommissioning. Three Mile Island Unit 1, shut down in 2019 as uneconomic, is being brought back to power artificial intelligence. Nobody saw that coming either.

The question is never "what will happen to Southern?" It is "what happens to me if I am wrong about Southern?"

Next step

Let's find your single point of failure

A Portfolio Preparedness Review is a straightforward exercise: what percentage of your net worth sits in one employer, what a bad decade does to your withdrawal plan, and what an unwind actually costs after tax. No product. No commission.

Wilder Bailey

Founder & Principal

Bailey Financial Services, Inc.

Watkinsville, Georgia

Wilder@BaileyFS.net

Contact

Bailey Financial Services, Inc. is a state-registered investment adviser. Information on this page is educational and general in nature and is not investment, tax, or legal advice, nor a recommendation to buy or sell any security. References to specific companies, projects, or historical events are illustrative only and are not an endorsement, a forecast, or a prediction of future results. Past performance does not guarantee future results. All investing involves risk, including loss of principal. Advisory services are offered only to clients or prospective clients where Bailey Financial Services, Inc. and its representatives are properly registered or exempt from registration.