Berkshire Stopped Selling — and Bought Houses, Airlines and a Department Store

For fourteen consecutive quarters, Berkshire Hathaway sold more stock than it bought. The cash pile climbed toward $400 billion and became its own kind of market commentary: the most patient capital allocator in America could not find anything worth owning. Then the Q2 2026 13F landed on the August 14 deadline and the streak was over. Berkshire was a net buyer for the first time in more than three years, and cash fell from a record $397.4 billion at the end of March to $365.5 billion at the end of June.

The headline went to Alphabet — 48.1 million shares added, roughly $17 billion, pushing it to about 9.4% of a $299.3 billion equity book and into the top five alongside Apple, American Express, Coca-Cola and Bank of America. That is the number everyone quoted. It is also the least interesting thing in the filing.

The rest of the shopping list

Strip out Alphabet and look at what else moved. A new position in D.R. Horton. An addition of 3.1 million shares to Lennar. Delta Air Lines up 17.5 million shares. Macy's up 4.3 million. The New York Times up 553,000. Constellation Brands sold out entirely, about eighteen months after it was bought. Trims in Bank of America, Kroger, DaVita, Ally Financial and Capital One.

Two homebuilders, an airline, a department store and a newspaper. In a quarter when the marginal institutional dollar was going into semis, hyperscalers and power infrastructure, Berkshire spent its first real buying quarter since 2023 on the domestic consumer and the American house.

Those positions are small relative to the book — Macy's and the NYT barely register next to a 22% Apple weight — which is exactly why they are worth reading. Small positions in a portfolio this size are usually the deputies, Todd Combs and Ted Weschler, expressing a view without needing it to move the needle. Big positions are the house view. When both are pointing at the same macro conclusion, the signal is stronger than either alone.

What the housing bet requires to work

Lennar plus a fresh D.R. Horton stake is a rate bet dressed as a stock bet. Homebuilders have spent two years absorbing mortgage rates that priced out the entry-level buyer, running incentives and rate buydowns that quietly ate margin. The equity works if financing costs come down and the structural shortage of housing units reasserts itself. It does not work if the Fed is stuck.

Which is the uncomfortable part. Those purchases happened by June 30. Since then, July PCE printed at 3.7% against a 3.6% consensus, a Fed official floated openness to hikes, and Kevin Warsh used Jackson Hole to talk about getting inflation back to target rather than about cutting. A rate-sensitive bet made in the second quarter is facing a materially less friendly third quarter. This is the standing hazard of 13F reading, and we wrote about it in the energy context: the filing is a photograph of June 30, published six weeks late, describing decisions made at prices that may no longer exist.

Berkshire is unusually forgiving of that lag, though. It does not trade around quarters. A homebuilder stake initiated in Q2 is meant to be held through whatever the Fed does in Q4.

The contrast trade

The late-August tape made the divergence vivid. Nvidia's guidance beat, semis ripped, and technology was the only S&P sector to finish green while healthcare, utilities and staples all sold off. Money rolled straight back into the crowded end of the market. Berkshire, as of its last disclosed positioning, owns almost none of that: no Nvidia, no Broadcom, no AI infrastructure complex beyond Alphabet — and Alphabet is arguably a search-and-cash-flow business that happens to own AI assets rather than an AI trade.

Run Berkshire through the overlap tool and the shape is unusual. Its largest positions are shared with almost no one in the top-fund universe. The names in the Q2 additions — Macy's, D.R. Horton, Delta — are held by so few of the tracked managers that they barely produce a consensus reading at all. That is not evidence they are right. It is evidence that if they are right, the payoff is not already in the price.

Reading the cash number properly

The $365.5 billion still sitting in T-bills is the more honest summary than any single purchase. Berkshire deployed roughly $32 billion of a $400 billion hoard and stayed overwhelmingly in cash. This is not a capitulation into risk. It is a firm that found a handful of things cheap enough to own while keeping most of its powder dry — which is a different macro statement than either "Buffett is bullish" or "Buffett is scared," and closer to what our macro consensus view picks up across the broader filer set: rotation at the margin, defensiveness at the core.

You can track the full position history on the Berkshire investor page as Q3 filings arrive in November.

---

Data sourced from public SEC 13F filings. Educational research only — not investment advice.