Looking back at Warren Buffett's five best and worst bets
Warren Buffett turns 95 this week, and Wall Street is still cribbing his homework. He’s the rare investor whose annual shareholder letter is quoted like scripture and whose portfolio is tracked almost like an index in its own right. He’s the patron saint of patience who built a mythology out of plain talk and compound interest. He’s the collector of parables who transformed a modest candy shop into a parable about moats, the crisis firefighter whose endorsement alone steadied Goldman Sachs and GE, and the disciplined allocator who confesses — in print — when he’s blown billions on a bad bet.
At 95, the scoreboard is lopsided in his favor, but his misses are as instructive as his hits. The Oracle of Omaha has long been caricatured as a Midwestern Luddite, wary of anything that looked like tech. In 1999, he warned Sun Valley’s moguls that the stock market had built a house of cards and refused to buy into companies he couldn’t explain on a napkin. Seventeen years later, he piled tens of billions into Apple, treating it not as a gadget maker but as a consumer-products behemoth with brand devotion to rival Coke. That single decision produced a profit measured in the tens of billions — and gave Berkshire Hathaway its largest holding ever.
His crisis-era performance was just as telling. In 2008 and again in 2011, while Wall Street bankers scrambled for lifelines, Buffett didn’t just sell capital. He sold confidence. Berkshire’s checks came bundled with double-digit coupons and stock warrants — sweetheart deals no one else could demand. Goldman Sachs, General Electric, and Bank of America effectively paid Buffett to bless them with solvency. The math was obscene — hundreds of millions a year in dividends, redemption bonuses, and billions more in equity gains when the storm passed. Yet Buffett insisted it was simple: He knew their franchises would survive, and he was happy to be paid like a king to wait.
And then there’s See’s Candies, the little West Coast chocolatier he bought in 1972 for $25 million that went on to generate more than $2 billion in profits with barely any new capital. Buffett has said he nearly walked away over a $5 million price gap — which would have been one of the most expensive examples of penny-wise, pound-foolish in corporate history. See’s wasn’t just a business; it was a revelation. It taught him to prize brand, pricing power, and low capital intensity — lessons that echoed through his biggest wins decades later.
That’s the Buffett method: parables written in cash flow. But the story isn’t one of unbroken triumph. Among the Cokes and Apples, though, there’s a Dexter Shoe or a USAir — deals that looked fine on the page and aged like milk. Buffett doesn’t hide them; he autopsies them, so shareholders and successors can see where discipline lapsed. To mark his 95th, here’s an honest ledger: five bets that made Berkshire a colossus and five that remind everyone that even the greatest investor of his generation pays tuition to the market.
Buffett didn’t touch Silicon Valley in the ’90s, then quietly began buying Apple stock in 2016, ultimately amassing a stake that cost roughly $30–$40 billion across several years. At its 2023–24 peak, Berkshire’s Apple position was valued north of $150 billion, before Buffett sold down a sizable chunk in 2024 and continued trimming in 2025 — moves he framed around prudence, concentration risk, and taxes. Even with the diet, Apple was designed to do what Buffett has always prized: ship cash to Omaha while compounding per-share economics inside Cupertino. Over the life of the trade, Berkshire has harvested many billions in cumulative gains — plus an estimated several billion more in dividends. The lesson isn’t that Buffett finally “got” tech; it’s that he recognized Apple’s moat looked like a consumer franchise with subscription-like behavior and extraordinary pricing power. That’s why even with a slimmer position, Apple still anchors Berkshire’s portfolio.
Berkshire owns about 151.6 million AmEx shares, a stake of roughly 21% acquired for just $1.3 billion in the early-to-mid 1990s. In his 2023 shareholder letter, Buffett noted that Berkshire’s share of AmEx’s 2023 earnings alone “considerably exceeded” that entire $1.3 billion cost — an almost impudent way to measure payback. The position has long sat in the inner circle of “we’ll likely own these indefinitely.” Buybacks at AmEx have quietly increased Berkshire’s ownership without Berkshire having to lift a finger. That is exactly how Buffett likes his compounding: automatic, tax-efficient, and low-drama.
Because the cost basis is so low, every incremental dividend is another reminder that time can be more powerful than timing. The stake has climbed into the $40-plus billion club in recent years, depending on market prices. It’s a case study in “buy right, sit tight.”
Coca-Cola is Buffett’s longest-held megacap and still a textbook of what he means by a “wonderful business.” Berkshire owns 400 million KO shares. At 2024’s rate ($1.94/share), those 400 million shares produced about $776 million in cash to Omaha. Coke has raised its dividend for more than six straight decades, and has already stepped it up again for 2025, lifting the payout to $2.04 for 2025, implying roughly $816 million this year. The position has been a master class in patience: The stock went sideways for years, but the dividend checks kept growing, and reinvested compounding did its work.
Coke also fits Buffett’s favorite economic profile: global brand, local bottlers (as in, not Coke itself) bearing the capital intensity, and immense distribution muscle. Over very long stretches, per-share value creation at Coke tends to show up in rising payouts and buybacks, not flashy acquisitions or moonshots. That’s fine by Berkshire, which hasn’t touched its 400 million shares in years — because why would you disturb a machine that reliably turns sugar water into cash?
In August 2011, as anxiety about Bank of America’s capital swirled, Buffett phoned CEO Brian Moynihan and offered not just money but validation. Berkshire invested $5 billion in 6% preferred stock — paying $300 million in annual dividends — plus warrants to buy 700 million common shares at $7.14. Six years later, after a dividend hike made it attractive to convert, Berkshire swapped the preferred into those 700 million shares, locking in what was then an instant double-digit-billion paper gain. In the years since, Berkshire added to the common, then trimmed in 2024 as prices surged; the stake remains one of Berkshire’s largest.
The broader lesson is reputational leverage. When Berkshire shows up with capital, the market reads it as due diligence you can’t buy at retail. Buffett knows that — and negotiates accordingly. The BofA package is exactly the kind of asymmetric structure Berkshire was built to write.
Buffett bought See’s in 1972 for $25 million. By 2007, See’s had delivered $1.35 billion in cumulative pre-tax profits to Berkshire — after requiring just $32 million of incremental capital to grow — because the business sells for cash, carries minimal inventories, and raises prices a little each year on a product people give to people they love. It is the platonic ideal of a Buffett company: a niche brand with irrational customer loyalty and a moat built from taste, tradition, and location. Buffett once wrote that See’s “has given birth to multiple new streams of cash” for Berkshire; the candy paid for other businesses. That’s compounded cash flow in its purest form.
And it’s a pretty sweet deal. See’s doubled its poundage over decades, not years, yet profits multiplied more than fifteenfold. The capital efficiency (earnings with “virtually no” reinvestment) is the secret sauce. Buffett has been candid that he almost blew the deal by haggling over $5 million. Imagine the counterfactual.
In 1993, Buffett bought Dexter Shoe, a proudly Made-in-Maine manufacturer with a sterling record, paying with Berkshire stock, not cash. The competitive moat dissolved under import pressure, the profits vanished, and the equity paid out — 25,203 A shares — became, in Buffett’s own words, a “most gruesome” mistake because those Berkshire shares would later be worth staggering sums. (Roughly $18.3 billion at today’s Berkshire A share price.) He has repeatedly called Dexter his worst deal and has been flagging it for years as a warning about overpaying for fragile moats and, worse, funding the purchase with your own undervalued stock.
Dexter’s failure wasn’t a random storm. It was a case study in globalization crushing a domestic cost structure that looked fine in backward-looking data. Buffett saw a great past; he misjudged the future industrial landscape. The lesson hardened one of his core rules: minimize stock issuance because dilution lingers far longer than acquisition euphoria. When it comes to Berkshire’s aversion to issuing shares, Dexter is Exhibit A.
In 2016, Berkshire bought Precision Castparts for $37 billion, betting on world-class aerospace components and a long runway of demand. Four years later, the COVID-19 pandemic detonated airframe production and supply chains; in 2020, Berkshire recorded a $9.8 billion impairment on PCC. Buffett said plainly: “I paid too much.” That’s not a thesis collapse so much as a price discipline failure — the kind he rarely commits and never forgets.
Strategically, PCC still fits Berkshire’s industrial spine. But mosaic moats — supplier relationships, certifications, switching costs — don’t immunize you from cycle and concentration. Overpaying for a very good business is still overpaying.
Buffett began buying IBM shares around 2011, praising its roadmap and client stickiness. Then he began selling those shares in 2017 after conceding he had “revalued it somewhat downward.” By 2018, he was out. At one point, Berkshire owned roughly 81 million shares, bought for about $13.8 billion; when the selling started, the stake was near break-even. The mistake wasn’t owning a bad company; it was misreading how quickly cloud-first competition and shifting enterprise spending would erode IBM’s advantage.
IBM taught a different lesson than Apple. In consumer tech, Buffett found habit and ecosystem; in enterprise tech, he found contracts vulnerable to platform shifts. It was a graceful retreat, but it was a retreat nonetheless, and it was a reminder that brand equity and durable pricing power aren’t synonyms. The former can be a story; the latter must be a number.
Buffett first learned the airline lesson in 1989, buying USAir preferred only to watch the dividend get suspended as the business “went into a tailspin.” He later got lucky — selling in a burst of industry optimism for a gain — but he immortalized the category in his 2007 shareholder letter: If a farsighted capitalist had been at Kitty Hawk, he wrote, “he would have done his successors a huge favor by shooting Orville down.” The industry devours capital, and for decades, the returns rarely justified the hunger.
Then came the contrarian detour: Berkshire reentered airlines in 2016, buying big stakes in the major ones, only to bail in April 2020 as the COVID-19 pandemic vaporized demand. Buffett told shareholders “the world has changed,” admitting Berkshire took losses on a position that had reached $7–$8 billion at cost. If USAir was the early warning, 2020 was the exclamation point: structurally tough industries can hand you long stretches of OK results, right up until they hand you a catastrophe.
In his 2008 shareholder letter, Buffett called out his own “terrible timing” on ConocoPhillips. He bought heavily when oil prices were high, then watched both crude and the stock fall. He wrote that the mistake had already cost Berkshire “several billion dollars.” As ever, he didn’t blame the commodity; he blamed his timing and the decision to let a macro view steer a micro allocation. Buying cyclical earnings at peak margins seldom ends well. He knew that; he did it anyway; he told you so in print.
Berkshire trimmed, moved on, and recycled capital into businesses with more controllable economics. The discipline to exit is a moat, too.
As Buffett prepares to hang up his crystal ball and hand Berkshire’s reins to coming CEO Greg Abel, the Oracle of Omaha’s coda is neat. Buffett dodged the late-’90s tech wreck because he didn’t pretend to understand it, then crushed it later by reframing one tech giant as a consumer staple. He steadied the system in 2008–11 and got paid like a king for doing so. He bought a box of chocolates that financed an industrial conglomerate, and he bought a shoemaker that torched a small nation’s GDP worth of Berkshire stock.
The common thread isn’t perfection; it’s temperament — knowing what you own, what you don’t, and when the price embeds too much hope. And at 95, that temperament is still compounding.