Berkshire's First Buyback in Three Years Was Not Buffett's Call

Berkshire Hathaway ended March with roughly $400 billion in the till. It ended June with $365.5 billion. The $34.5 billion difference is the first meaningful drawdown in the company's sixty-year history that Warren Buffett did not authorize.

The man who did is Greg Abel. He took the chief executive's chair in January 2026; Buffett stayed on as chairman. When Berkshire published second-quarter results on Saturday, August 8, the question the market had been asking for months finally got an answer with a number attached: what will the new management do with all that money?

The answer came in three lines: $4.5 billion of share repurchases, $19.8 billion of net stock purchases, and $10 billion committed to Alphabet's equity raise for AI infrastructure.

How the Pile Got This Big

The cash accumulation was never a strategy. It was the residue of an absence. In his final years running the company Buffett sold stock — including most of the Apple position — and did not replace it with anything. He also stopped buying back Berkshire's own shares. The company's repurchase rule is not a written commitment but a measuring stick: buy when the shares trade below management's conservative estimate of intrinsic value, do not buy when they trade above it.

For three years cash grew in the shadow of that rule. Today's $365.5 billion equals roughly a third of the company's approximately $1.1 trillion market capitalization.

Abel restarted the program in March, saying the stock was trading below intrinsic value. The first quarter's purchases came to only $234 million. The second quarter's came to $4.5 billion. The six-month total is $4.8 billion.

The Trouble With "Profit Doubled"

Most of Saturday's headlines led with net earnings doubling: $12.4 billion to $25.7 billion. That is true, and it tells you almost nothing about the business.

An accounting rule in force since 2018 pushes daily portfolio price swings straight into the income statement. Because Berkshire's equity book runs to several hundred billion dollars, a move of a few percent in that book can generate more reported "profit" than the railroad and the utilities earn in a quarter. Buffett spent years telling shareholders to ignore that line. On Saturday that line was the headline.

Operating earnings carry an asterisk of their own. About $1.2 billion of the increase came from currency moves on yen- and euro-denominated debt. Strip that out and the underlying businesses grew around 5%, not 16%.

The Mechanism: When a Buyback Creates Value

A repurchase is a company buying its own shares in the market and cancelling them. The share count falls; the same company is divided into fewer pieces. Nothing about that automatically means shareholders came out ahead. Whether they gained or lost depends entirely on the price paid.

At Berkshire's scale the practical consequence of that arithmetic is surprising. $4.5 billion is about four-tenths of one percent of a $1.1 trillion market value. Even if the shares were bought a full 10% below intrinsic value, the gain to remaining holders is measured in hundredths of a percent.

So the buyback's monetary effect is small and its informational value is large. Remember the rule: management buys only when it considers the stock cheap. The $4.5 billion is a statement about what Abel thinks the shares are worth — and so, for three years, was the absence of any statement at all. In a valuation argument, the answer management gives with its own money outweighs the sentences in the presentation.

Cash Has a Price

To understand the timing, look at what the cash earns. Berkshire's pile sits largely in short-dated Treasury bills, so its yield tracks short-term rates. On $365.5 billion, every one-percentage-point move in those rates is worth roughly $3.7 billion a year in income — with no decision taken, simply because the curve shifted.

The effect showed up this quarter: insurance investment income fell 9.1% to $3.1 billion, which the company attributed largely to lower short-term rates.

The macro backdrop is softening too. The July employment report released on August 7 showed nonfarm payrolls falling by 23,000 against expectations of an 86,000 gain, with May and June revised down by a combined 103,000. The 10-year Treasury yield slipped to 4.60%. More: Interest Rates and Bonds and Employment Data

Where the Money Went

The quarter's most striking move landed on June 1. Alphabet announced an $80 billion equity raise to fund AI infrastructure: $30 billion through underwritten public offerings, $40 billion via an at-the-market program starting in the third quarter, and $10 billion placed directly with Berkshire. Berkshire took $5 billion of Class A stock at $351.81 a share and $5 billion of Class C stock at $348.20. Per Forbes, Buffett negotiated the deal himself.

Alphabet guides to capital expenditure of $180–190 billion for 2026 and says 2027 will increase significantly from there. Sundar Pichai's reasoning is short: "The risk of under-investing is dramatically greater than the risk of over-investing." Berkshire has therefore tied part of its cash to the largest capital-spending wave outside its own operating businesses.

The window above spans from the placement in early June through to the present.

The Alphabet position first appeared in the third quarter of 2025, tripled during Abel's first quarter, and grew again through June's private placement. At roughly $41 billion it is now the portfolio's fourth-largest holding, behind only Apple and American Express. In that same first quarter the Visa, Mastercard, Domino's Pizza and Amazon positions were closed and about $2.6 billion went into Delta. For a portfolio built on the principle of not buying what you don't understand, AI infrastructure moving this close to the center is the clearest evidence of a change in style.

Timeline

What the Businesses Did

The portfolio moves took the headlines, but Berkshire's body is still the operating companies. The picture there is mixed.

Insurance is the weak link. GEICO's loss ratio deteriorated by roughly five points to 76.6%: claim frequency rose, average bodily-injury severity climbed 10–12%, and commission and advertising costs added to the strain. Pretax underwriting earnings fell from $1.821 billion to $994 million.

BNSF, by contrast, moved 6.5% more carloads and grew earnings 6.3% despite a 68% jump in fuel expense. Energy earnings rose 26.9% on a 3.1% increase in retail electricity volumes. The data-center power debate reaches Berkshire's income statement through exactly that line.

The chart above shows the market's general path; Berkshire's own shares lagged it. Through August 7 the stock was up 3.5% for the year against 14.1% for the S&P 500.

The Other Side

What Remains

The real question is still open: will $365 billion turn into an acquisition large enough to matter at this scale? The $34.5 billion spent in a quarter shrank the pile by only 8.6%.

A second clock is running on the ownership side. In July Buffett transferred roughly $6 billion of Class B stock to four family foundations and said he aims to distribute all of his Berkshire shares within about eight years. That means his voting power disperses over time as well. Buffett's shadow currently balances Abel's decisions; that shadow shrinks with the calendar.

This piece draws on Berkshire Hathaway's second-quarter results of August 8, 2026, Alphabet's June 1, 2026 equity-raise announcement, and reporting on the quarter from Forbes, Fortune and the Associated Press. Segment comparisons and the separation of currency-driven gains come from independent analyst notes and may not map exactly onto the company's official segment definitions.