Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
One appealing idea is spreading through the AI bear camp. The very rich have borrowed heavily against their soaring tech shares, the argument goes, so when the bubble bursts the margin calls will land on them first. On this view a crash would be a great leveller, the moment the billionaire class finally gets a bill it cannot pay.
Part of this is true. A serious AI crash would destroy some enormous fortunes, and borrowing against stock would decide which ones. But the wider conclusion, that the rich as a group come out of a crash poorer relative to everyone else, runs against almost every modern market cycle. Unless something unusual happens this time, the likelier outcome is a handful of spectacular personal collapses while the class as a whole comes out of the wreckage holding an even larger share of what is left.
The interesting question is not whether the rich lose money. They will. It is which kinds of wealth survive a crash and which kinds disappear.
The Uncomfortable Arithmetic of Owning Most of the Market
Start with who actually holds the exposure. According to the Federal Reserve’s distributional accounts, the top 1% of US households own 50.9% of all corporate equities and mutual fund shares, the highest reading in the series. The top tenth of households holds most of the rest.
That has an obvious consequence. Measured in dollars, nobody loses more in an equity crash than the people who own the equities. When technology stocks fall by a third, the headlines fill with billionaires losing tens of billions in a week, and those figures are real.
[Also See: The Real History of Meny and Who Controls You]
But a paper loss and a permanent loss are different things. A household that owns shares outright, owes nothing against them and has no need to sell can watch its net worth halve and still own exactly as many shares as before. When prices recover, so does the fortune. What turns a paper loss into a permanent one is being forced to sell near the bottom, and that is where borrowing comes in.
Borrow, Don’t Sell: The Machinery Behind Paper Fortunes
Among the very wealthy, borrowing against shares is less about leverage than about tax. Selling appreciated stock triggers capital gains. Borrowing against it does not. So founders and large holders take out loans secured on their shares to fund homes, philanthropy, other ventures or just living expenses, and they keep the stock. The strategy is popular enough to have a nickname: buy, borrow, die.
The mechanics matter. Regulators warn that with a securities-backed credit line, a fall in collateral value can trigger a maintenance call that must be met within days, and the lender can sell the securities, often without notice. The same guidance singles out portfolios concentrated in one stock or sector as especially exposed, because a single shock can force a sale close to the low.
The amounts at the top can be very large. Oracle’s latest proxy statement shows that Larry Ellison has pledged about 413 million Oracle shares, roughly 36% of his holding, as collateral for personal loans, up 67 million shares on a year earlier, at a time when Oracle stock had fallen by roughly half over twelve months. Press reports say he has also pledged shares in other companies this year. Oracle’s board states that he can repay the loans without touching the pledged shares, and nothing suggests otherwise. The example matters not because anyone is in trouble, but because it shows how much of the world’s largest fortunes is tied, through collateral, to the price of a single AI-exposed stock.
What Forced Selling Looks Like in Practice
History offers clear examples of what happens when that link breaks.
In October 2008, as natural gas prices and energy shares collapsed, Chesapeake Energy chief executive Aubrey McClendon was forced to sell substantially all of his roughly 33 million Chesapeake shares over three days to meet margin calls. One of the best known founders in American energy lost most of his stake in his own company within a week, not because the business had failed but because his loans were secured on its share price.
The same month, Sumner Redstone’s holding company sold about $233 million of nonvoting Viacom and CBS stock. The loan in question was reportedly not secured directly on those shares, but its covenants were tied to the share prices of the two companies, so falling prices pushed it into default anyway. Leverage does not have to take the form of a classic margin loan to force a sale.
The most dramatic modern case came in March 2021. Bill Hwang’s family office, Archegos, had built enormous concentrated positions through derivatives that let it borrow heavily from several banks at once. When a few of those stocks fell, the margin calls could not be met, the banks dumped the shares, and reported losses for Hwang ran into the tens of billions within days while Credit Suisse alone lost about $5.5 billion. Hwang was later convicted of fraud.
The pattern across all three is the same: concentration, leverage, and collateral whose value depended on the same asset that was falling. An AI crash would almost certainly produce cases like these, most likely among founders, early employees and family offices whose wealth sits in one or two AI names with loans stacked on top.
Why the Class Usually Wins Even When Individuals Lose
Now look at the bigger picture. If crashes were great levellers, the share of wealth held by the top would fall sharply after each one and stay down. The data shows a brief dip followed by a rebound to new highs.
In the Federal Reserve’s own analysis of its distributional data, the top 1% share of US net worth slipped from about 29.9% in 2007 to a low of about 27.8% in 2009, then climbed to around 31.7% by 2016. The same thing happened in 2020. The share fell to about 29.0% during the pandemic crash and has since risen to a record 32.5% in the second quarter of 2026.
Meanwhile, the middle of the distribution took a far deeper and longer hit after 2008. The Fed’s Survey of Consumer Finances found that inflation-adjusted median family net worth fell 38.8% between 2007 and 2010, driven mainly by the collapse in house prices. Typical households held their wealth in a single leveraged asset, their home, and many lost jobs at the same time.
Three forces explain the asymmetry. First, most large fortunes are not forced sellers. They own shares outright or borrow modestly against them, so they can wait. Second, they hold liquidity when it is scarce. Cash, bonds and credit lines let them buy shares, property and distressed businesses at crisis prices, which is exactly when long term returns are made. Third, the policy response after a crash tends to lift asset prices first. Rate cuts and central bank asset purchases restored stock and bond prices years before wages and employment fully recovered, and those who owned the assets captured the rebound.
The Dot-Com Exception, and Why It Matters Now
There is one important exception, and it is the one most relevant to AI. After the March 2000 peak, the Nasdaq did not quickly bounce back. It took more than 15 years for the index to close above its dot-com record. Many of the paper billionaires of that era never recovered at all, because their wealth was not diversified capital. It was a single stake in a single overvalued company, and when that company shrank or vanished, so did the fortune.
The 2008 and 2020 crashes were broad shocks followed by aggressive rescues. The dot-com crash was a sector bust, and the sector’s valuations never returned for most of the companies involved. If an AI crash looks more like 2000 than 2008, the recovery that rescues the diversified rich may not arrive for the concentrated rich. Holding the right asset class does not help if you hold the wrong company.
Three Kinds of Rich in an AI Crash
It helps to separate the wealthy into three rough groups rather than treating them as one class.
The first is diversified money: broad equity portfolios, property, private businesses outside technology, bonds and cash. This group suffers large paper losses, rarely faces forced selling and is usually among the biggest buyers near the bottom. History suggests it comes out relatively stronger.
The second is concentrated founder wealth: fortunes tied to one AI company, often with substantial loans secured on those shares. Most of these holders can survive a sharp fall, especially when lenders and boards are comfortable, as Oracle’s is. But a deep and lasting decline in a single stock is exactly the scenario where pledged positions turn into forced sales, and a deep fall in one stock can wipe out a large share of one of these fortunes.
The third is leveraged speculation: family offices, funds and individuals running borrowed exposure through margin loans, options or swaps on AI names. This is where the Archegos-style blowups come from, and where losses can be total.
What This Means for Ordinary Investors
For anyone outside these groups, the lessons are practical rather than moral.
Do not build a strategy around the rich being forced to sell. Most of them will not have to, and the few who do will create sharp but temporary dislocations rather than a lasting collapse in prices. Those forced sales are opportunities for patient buyers, not proof that the whole market is finished.
Avoid becoming the forced seller yourself. The households that suffered most in past crashes were the ones carrying leverage on a single asset and losing income at the same moment. Margin debt on concentrated tech positions turns a recoverable drawdown into a permanent loss.
Treat liquidity as an asset with its own return. Cash and short term bonds look dull at the top of a cycle. Their value shows up at the bottom, when they are the only thing that lets you buy what others must sell.
And if you want an early warning signal, read the proxy statements. Share pledging by insiders must be disclosed, and rising pledges at heavily valued companies tell you where forced selling could come from if prices keep falling.
The comforting story says an AI crash would humble the people at the top. The record says it would humble some of them, very publicly, and leave the rest holding more of the market than before. The difference between the two groups is not how much they own. It is whether anyone else can force them to sell.
