Berkshire Hathaway added more than $17 billion of Alphabet in the second quarter, and CNBC reported that the position is now the conglomerate’s third-largest holding at roughly $36.6 billion, behind only Apple and American Express. The headline everyone ran was that Berkshire loves Google. The more interesting fact is buried in how the shares were acquired.
Roughly 60% of the 48.1 million shares Berkshire picked up during the quarter did not come off an exchange. They came directly from Alphabet, in a $10 billion private placement the two companies announced in early June. Berkshire bought around $7 billion on the open market. The rest was a financing transaction.
Berkshire Is Not Just Long Google, It Is Funding the Buildout
That distinction changes what the trade means. Buying stock on the open market is an opinion about price. Taking a $10 billion private placement straight from the issuer is something closer to underwriting, and the thing being underwritten is AI capital expenditure.
We covered the context when Alphabet went to the market: the company raised $80 billion in equity to fund AI infrastructure, and the stock dropped on the news while Berkshire stepped in as an anchor. That is the same pattern Berkshire has run for a century, from Goldman Sachs in 2008 to Occidental in 2019. Berkshire shows up when a large issuer needs size, certainty and a name that calms other holders, and it gets paid for supplying those three things rather than for being clever about entry price.
Which reframes the position. Berkshire is not primarily betting that Alphabet wins the model race. It is renting out its balance sheet to the largest capital raise in the sector, at terms negotiated privately, at a moment when the market was punishing that raise. The Alphabet exposure is real, but the structure is a financing business wearing an equity position’s clothes.
Warren Buffett told CNBC in June that the original Alphabet purchase, begun in the third quarter of 2025, was his idea. The scaling of it into a private placement is a different decision, made under different management, and it deserves to be read separately.
Abel Ended a 14-Quarter Selling Streak in One Print
The portfolio-level number is the one that should move people’s models. Berkshire was a net buyer of $19.8 billion of equities in the second quarter, ending 14 consecutive quarters of net selling. That is the largest net-buying reversal in three and a half years, and it is the first full quarter that reads unambiguously as Greg Abel’s. Forbes framed the filing as Berkshire turning buyer, which understates it slightly: the conglomerate did not drift into buying, it reversed a posture it had held since 2022.
For most of the Buffett endgame, Berkshire’s message was that nothing was cheap. Cash piled up, positions got trimmed, and the implied view was that the patient thing to do was wait. A $19.8 billion quarter says that view has changed, or that the person holding it has.
The rest of the filing fills in the shape of it, and Yahoo Finance’s read of the adds and the bank trims captures the rotation:
- Increased: Alphabet, Delta Air Lines, Lennar, Macy’s and the New York Times
- Trimmed: Bank of America, DaVita, Kroger, Ally Financial and Capital One
- Exited: Constellation Brands
Read the two columns together and the rotation is legible. Money came out of consumer lenders and regional banking exposure and went into a technology platform, a homebuilder, an airline and two consumer names. It is not a defensive posture. Abel has also shown he will write large private checks elsewhere, including the $6.8 billion Taylor Morrison acquisition in housing, which pairs neatly with adding Lennar.
The Question the Filing Does Not Answer
A 13F is a rear-view mirror. It reports positions as of June 30 and lands six weeks later, so the quarter’s prices are gone and the reasoning is never disclosed. What we get is the fact of the trade.
The fact here is that the most price-sensitive institution in American finance stopped waiting, and that the single largest thing it did was supply capital to an AI infrastructure buildout at a moment when public investors were selling that buildout. Alphabet had already earned its way into the Dow and past a $52,000 share price with Berkshire’s earlier $10 billion in the story, so this is an escalation of an existing view rather than a new one.
Michael Burry, who has spent the year warning that AI capital spending is being depreciated over unrealistically long schedules, sits on the other side of that argument. If he is right about the accounting, the companies raising tens of billions to buy compute are overstating earnings, and the institutions financing those raises are funding an asset base that ages faster than the books admit.
Berkshire has now taken a public position on that question, for $17 billion, and about $10 billion of it was placed where no other buyer had to agree on the price. The next few quarters of Alphabet’s capex disclosures will settle whether that was underwriting or catching something on the way down.