Buffett's Most Expensive Mistake Doesn't Appear on Any Balance Sheet

On November 7, 1993, Warren Buffett bought a New England shoemaker named Dexter and paid for it with 25,203 of Berkshire's most precious Class A shares — then worth $433 million.¹ In that year's letter to shareholders, he called it 'one of the best-managed companies Charlie and I have seen.'¹ He was wrong on both counts, and the way he was wrong has become the most quoted mistake in the history of investing. But the number everyone quotes is the wrong number — and the mistake everyone quotes is the wrong mistake.

The official story is that Dexter Shoe was a $433 million blunder. That is precisely backwards. Buffett didn't lose $433 million; he handed over a sliver of the single best-compounding stock on the planet to buy a business that was about to be erased by imports. The loss was never the cash he paid. It was the cash those shares would have become.

Why the loss number keeps getting bigger every year

Here is the part that makes Dexter genuinely instructive rather than merely embarrassing. Because Buffett paid in stock and not in cash, the cost of the mistake is not fixed — it tracks Berkshire's share price, and Berkshire's share price has done nothing but climb. In his 2007 letter he called Dexter 'the worst deal that I've made' and put the cost to shareholders at $3.5 billion.² Seven years later, in 2014, he updated that same surrendered stake: 'the shares I used for the purchase are now worth about $5.7 billion,' and reached for darker language — 'the most gruesome' mistake, one that 'deserves a spot in the Guinness Book of World Records.'³ By 2024, those same 25,203 shares were valued at approximately $14.4 billion — a figure that keeps climbing with Berkshire's price.⁹ There is no canonical loss number for Dexter. There is only a loss that compounds for as long as Berkshire does.

This is the mechanism beneath the headline. When you pay cash for a bad asset, the damage is capped at the cash. When you pay with a security that compounds at extraordinary rates, every dollar of purchase price keeps multiplying inside the seller's pocket and inside your own opportunity cost forever. Dexter wasn't a $433 million wound. It was a $433 million seed planted in the wrong field — and the field grew anyway, just not for Berkshire.

The far bigger misses that never showed up as losses

Dexter at least had a price tag, however elastic. Buffett's truly expensive errors had none — because you cannot write down a stock you never bought. At the 2018 annual meeting he said it flatly: 'I made the wrong decisions on Google and Amazon.'⁵ On Google the indictment is almost comic: Berkshire's own Geico was paying Google heavily for ads around the time of its IPO — Buffett was watching the cash register ring from inside the store — and he still didn't buy.⁵ On Amazon he admitted he had underestimated Jeff Bezos and the company's ability to achieve what he called a business 'miracle' across e-commerce and cloud.⁵ Neither omission ever cost Berkshire a reported dollar. Both cost it more than Dexter ever could.

Note the correction the popular narrative gets wrong: these were not permanent abstentions. Berkshire added Amazon in 2019 — a purchase made by one of its portfolio managers, not Buffett himself¹⁰ — and initiated an Alphabet position in Q3 2025.¹¹ The miss wasn't the company — it was the decade. Buffett didn't fail to recognize Google and Amazon as great businesses; he failed to act while they were small. The cost of an omission is the entire compounding curve you watched from the sidelines, and that curve is invisible to every accounting statement ever drafted.

How a confession becomes a competitive advantage

Now watch what Buffett does with all of this — because the disclosure is the strategy. The errors don't get buried in a footnote; they get top billing in the shareholder letter, in his own words, with the worst language he can find. The 2023 annual report carries his admission that Munger told him in 1965 that buying control of Berkshire's textile business itself was 'a dumb decision,' and that Munger had to teach him how to dig out of it.⁴ His self-described foundational error is literally the name on the building. The 2024 letter opened on mistakes again, and Buffett noted he had used the words 'mistake' or 'error' 16 times across his 2019–2023 letters.⁸ He even adopted Tom Murphy's rule — 'praise by name, criticize by category' — keeping company names out of the criticism while keeping the criticism in.⁸ The confession is engineered.

But didn't some of the 'mistakes' actually make money?

The honest objection is that the 'biggest mistakes' canon is sloppier than it pretends, and cuts against my own framing in places. Take USAir, a fixture of every Buffett-blunder list. It caused him years of anxiety and near-total write-downs — yet his 2007 letter records that Berkshire ultimately sold the common shares it received 'for a hefty gain' in 1998.² A 'mistake' that ends in a profit is a strange kind of disaster. The airlines are messier still: the popular telling makes the 2016 bet entirely Buffett's, but the initial American, Delta, and United positions were initiated by one of Berkshire's two portfolio managers; Buffett added Southwest in Q4 2016 and enlarged all four through 2019, holding roughly 10% of each carrier by that December.⁶⁷ When he dumped all four at a loss at the May 2020 meeting, he was candid about the cause: 'I was wrong about that business because of something that was not in any way the fault of four excellent CEOs.'⁷ COVID-19, not bad underwriting.

So the steelman holds halfway: the commission errors that fill the listicles are routinely overstated, mis-dated, or quietly profitable. But it actually strengthens the real point rather than weakening it. The misses we can count — Dexter, USAir, the airlines — are the small ones, and we still get their dollar figures wrong. The misses we cannot count — the years of Google and Amazon — are the enormous ones, and they leave no figure at all. The market is loud about the loss it can see and silent about the loss it can't, which is exactly backwards from where the money actually went.

Buffett's genius was never that he avoided mistakes; it was that he chose which ones to make legible. He turned a depreciating shoemaker into a permanent teaching case and a foundational textile blunder into the name on his door, while the real damage — the compounding he watched and didn't buy — stayed politely off the books. The lesson isn't 'don't make mistakes.' It's that the costliest error is the one with no entry to write down, and the smartest move is to confess the cheap ones so loudly that everyone forgets to add up the expensive ones.