Greg Abel spent the quarter deploying Warren Buffett’s cash pile. The Alphabet trade shows what Berkshire’s balance sheet is worth now, and it is worth far less than it was in 2008.
Berkshire Hathaway reported second-quarter 2026 net income of $25.67 billion, more than double the $12.37 billion it earned a year earlier, and its cash and Treasury holdings fell sequentially for the first time in about four years, to roughly $364.7 billion from a record $397.4 billion. The coverage read this as Greg Abel finally spending. The more useful read sits inside one line item: Berkshire bought $10 billion of Alphabet common stock in a private placement priced more than 6% below the market, which values the service Berkshire sells at roughly $640 million. In 2008 the same service cost Goldman Sachs a 10% perpetual preferred dividend plus warrants on $5 billion of stock. Berkshire still has the biggest checkbook in corporate America. The premium it can charge for using it has collapsed.
Berkshire has one asset that no competitor has copied: the ability to commit $10 billion in a phone call, with no financing condition, no syndicate, and no diligence period that leaks to the press. For fifteen years that capability priced like scarce goods. Companies in trouble paid Buffett a coupon, gave him a preference above the common, and threw in warrants on top.
Alphabet paid a one-time discount on ordinary shares. No coupon above the common dividend, no liquidation preference, no warrants. Then it went and raised another $70 billion from everyone else.
What happened
Berkshire released second-quarter results on Saturday, August 8. Net income reached $25.67 billion, or $17,868 per Class A share, against $12.37 billion, or $8,601, a year earlier. Most of that swing came from accounting: pre-tax investment gains of $16.1 billion, including $10.9 billion of unrealized marks on equity holdings that Berkshire has to run through net income under GAAP.
Operating earnings, the number Buffett always told shareholders to watch, rose 16.3% to $12.98 billion from $11.16 billion. Total revenue hit $101.8 billion, up 10%.
The segments split cleanly:
| Segment | Q2 2026 pre-tax earnings | Change |
|---|---|---|
| Manufacturing, service and retailing | $4.47B | +24% |
| BNSF Railway | $1.56B | +6% |
| Berkshire Hathaway Energy | $891M | +27% |
| Insurance underwriting | $1.73B | -13% |
| Insurance investment income | $3.06B | -9% |
Precision Castparts led on aerospace demand with pre-tax earnings up 34.2%. The service group gained 20.6% on TTI, NetJets and FlightSafety. GEICO dragged underwriting down. Clayton Homes, Abel’s own stated bellwether for housing, saw pre-tax earnings fall 3.5%.
On capital allocation, three moves mattered. Berkshire bought $23.5 billion of equities and sold $3.7 billion, ending fourteen consecutive quarters of net selling. It repurchased 478 Class A shares and more than 8 million Class B shares for about $4.5 billion, all of it in May and June, against $235 million in the entire first quarter. And it closed the $6.8 billion acquisition of homebuilder Taylor Morrison on July 24, which lands in the building products group next quarter.
Class A shares closed Friday at $780,086, up 3.4% for the year against a 13% gain for the S&P 500.
The backstory
The Alphabet trade started before Abel did. Buffett has said he initiated the position himself in the third quarter of 2025, and Berkshire kept adding through the first quarter of 2026.
The June 1 transaction was different in kind. Alphabet announced an $80 billion equity raise that day to fund AI compute: $30 billion in concurrent public offerings, a $40 billion at-the-market program, and a $10 billion private placement to Berkshire. Berkshire took $5 billion of Class A common at $351.81 per share and $5 billion of Class C at $348.20, both priced at a reported discount of more than 6% to Alphabet’s June 1 close.
Alphabet needed the money for a reason BMA covered three weeks ago: capital expenditure of $44.9 billion in a single quarter, full-year guidance raised to $195 to $205 billion, and free cash flow at negative $5.9 billion. A company with $242.5 billion in cash and securities was selling equity at a discount because its own operations no longer funded its buildout.
The plan
Abel’s deployment has a shape, and it runs in two directions.
Where Berkshire buys control of an operating business, it pays up. Taylor Morrison cost $72.50 a share in cash, a 24% premium to the May 29 close, valuing the builder near $8.5 billion including debt. Abel described the goal as combining Berkshire’s site-built homebuilding into one platform, which puts Taylor Morrison alongside Clayton Homes, Shaw, Benjamin Moore, Acme Brick, Johns Manville and MiTek. Berkshire already sells the flooring, the paint, the brick, the insulation and the connectors. Now it owns a national builder that buys all of them.
Where Berkshire supplies capital as a minority holder, it collects. That is the Alphabet trade, and it is the same trade Buffett ran on Goldman Sachs in 2008, on Bank of America in 2011 and on Occidental in 2019.
The difference is the price.
Goldman paid a 10% perpetual preferred dividend, about $500 million a year on $5 billion, plus five-year warrants on 43.5 million shares at $115. When Goldman redeemed in March 2011 it wrote a check for $5.64 billion and the warrants stayed live. Occidental paid 8% on $10 billion in 2019, plus warrants on 83.86 million shares at $59.62.
Alphabet paid roughly $640 million once, in the form of a discount, and owes Berkshire nothing further beyond the ordinary common dividend. On a per-dollar basis, Goldman’s terms cost more in fifteen months than Alphabet’s cost in total.
The business model angle
Berkshire’s $364.7 billion is not idle money waiting for a bargain. It is inventory for a business, and the business sells certainty of settlement.
Every large capital raise has the same problem. The issuer needs the money on a date, and the market needs to believe someone credible has already looked at the books. Underwriters solve the first half and charge fees. An anchor investor solves the second half and charges a discount. Berkshire has run that second business at higher margins than anyone because Buffett’s name did work no other balance sheet could do.
Look at what Alphabet bought. A $10 billion cornerstone order inside an $80 billion raise, announced in the same press release, from the most conservative capital allocator in the market, at a moment when investors were selling Alphabet stock over capex. The discount was the fee. Berkshire booked about $640 million of value at settlement for showing up.
Now look at what it cost. In 2008 Berkshire priced the same service as distress lending, with a coupon, a preference and equity upside on the side. In 2026 it priced as a block trade.
The gap measures competition, not generosity. Private credit funds now clear tens of billions in a quarter. Sovereign wealth funds write nine-figure checks without a committee. Microsoft, Nvidia and Alphabet itself deploy strategic capital at a scale that would have been unimaginable in 2008. When five other parties can wire $10 billion by Friday, speed stops being a product and becomes a commodity, and the seller loses pricing power.
The buyback tells the same story from the other end. Abel repurchased $4.5 billion of Berkshire stock in two months after buying almost none in the previous three. The best risk-adjusted use he found for that capital was his own company, in a quarter when he was also telling shareholders that public and private markets both look expensive.
The risk
The strongest objection to all of this is counterparty risk, and it deserves a hearing.
Goldman in September 2008 was two days removed from converting to a bank holding company and might not have existed by Christmas. Occidental in 2019 was fighting Chevron for Anadarko with a stretched balance sheet and oil in the seventies. A 10% coupon and a warrant package priced real probability of loss. Alphabet is a trillion-dollar business with $242.5 billion of liquid assets and a $514 billion cloud backlog. On a risk-adjusted basis, 6% for a week of settlement risk against a AA credit may be the better trade.
That objection is fair and it does not rescue the franchise. Berkshire earned outsized terms in 2008 because almost nobody else could act. Alphabet in 2026 had eighty billion dollars of demand for its paper. Berkshire got the same allocation any large institution would have taken, on terms any large institution would have accepted. The value of being Berkshire showed up in the press release, not in the term sheet.
Two more risks sit under the quarter. Buffett initiated the Alphabet position, which complicates the story of Abel putting his stamp on the company. And GEICO’s underwriting deterioration matters more than the headline suggests, because insurance float is the raw material that makes the whole certainty business possible. Investment income falling 9% on a near-record cash balance is a signal about yields, and the pile only justifies itself while Treasury bills pay.
Quick questions
Did Berkshire’s cash pile actually shrink? Yes. Cash and Treasury bills ended June at roughly $364.7 billion, down from the record $397.4 billion three months earlier. The decline understates deployment, because the company also generated about $13 billion of operating earnings during the quarter that flowed back in.
Why did net income double when operating earnings rose only 16%? GAAP forces Berkshire to run unrealized gains and losses on its equity portfolio through net income. Pre-tax investment gains hit $16.1 billion, including $10.9 billion of paper marks. Buffett has spent years telling shareholders to ignore that line.
Is Taylor Morrison in these numbers? No. The deal closed July 24, so it appears in the building products group starting next quarter. The $6.8 billion also came out of cash after the June 30 balance sheet date.
What does Berkshire own in Alphabet now? Reported estimates put the combined stake near $31 billion to $41 billion depending on the date, making Alphabet a top-five holding alongside Apple, American Express, Bank of America and Coca-Cola. The full 13F lands next week.
Is Berkshire done selling stocks? The quarter broke a fourteen-quarter streak of net selling with $23.5 billion of purchases against $3.7 billion of sales. One quarter is not a policy.
The Business Model Analyst Take
Every profile of Greg Abel asks the same question: what will he buy? That question assumes the cash pile is a shopping budget.
Treat it as inventory instead and the quarter reads differently. Berkshire’s product is unconditional capital delivered on demand, and Q2 gave you a price print. Alphabet paid roughly 6.4% once for a cornerstone order. Goldman paid 10% a year plus warrants. The service is the same. The market for it is not.
If you build businesses, the pattern is worth stealing and worth fearing. Berkshire spent fifty years turning a balance sheet into a brand, so that being the buyer became its own source of pricing power. That is the strongest kind of moat, because it charges rent on an asset that costs nothing to maintain. It is also the kind that erodes without a single bad decision. Nobody at Berkshire made a mistake. Capital simply got less scarce.
Abel’s real problem is not finding deals. It is that the thing his predecessor spent half a century making valuable has been commoditized by an ocean of private credit, and no amount of patience fixes a pricing problem.
Watch the terms, not the totals. If the next $10 billion Berkshire commits comes with a coupon and warrants attached, the franchise is intact. If it comes as another block at a single-digit discount, Berkshire is a very large index fund with a railroad.
