Boring just had its best week in a while. Berkshire Hathaway's B shares rose more than 5% while the S&P 500 fell nearly 2.7% and the Nasdaq Composite dropped 3.4%. Thursday's heavy tech selling pushed the Nasdaq more than 10% below its recent high, which meets the usual definition of a correction. A month ago Berkshire trailed the S&P by 13.6 percentage points for the year. It now leads by 5.8 points.
The swing shows what investors reach for when confidence in expensive growth stories wobbles. Berkshire has relatively little direct exposure to artificial intelligence. It also holds a large cash pile, $365.5 billion as of June 30, and it bought back $4.5 billion of its own stock in the second quarter. The shares trade at roughly 14 times trailing earnings. In a market worried that trillions in data-center spending won't earn their keep, a profitable conglomerate with an enormous cushion starts to look like shelter. The shares have now recovered all of the roughly 15% they lost after Warren Buffett said in 2025 that he would step down as CEO.
The rally has a wrinkle. Berkshire is not immune to the AI trade. It owns about 106 million Alphabet shares, including $10 billion bought directly from the company in June at roughly $350 a share. The stock now trades near $284. The position was worth $37.8 billion on June 30. It is now valued around $30.2 billion, which drops it behind Coca-Cola as Berkshire's fourth-largest holding. One Barron's columnist has called the stock-picking record of the current decade underwhelming. He points out that Berkshire sold much of its Apple stake near $200 a share and that Apple now trades above $340, and that the S&P 500 has returned about 150% since the end of 2019. Counterpoint: Apple is still the portfolio's largest holding at $78 billion, and it is up some 1,200% since Berkshire first bought it in 2016.
Smaller moves point the same way. Berkshire added about 924,000 shares of Sirius XM at an average of $29.10, a modest $27 million purchase on top of a $4.1 billion stake, after the company raised guidance and reported a 14% jump in free cash flow. And Berkshire's utility arm faces real pressure. PacifiCorp's Pacific Power is requesting a 57% rate increase in California, and Oregon wildfire litigation could expose it to enormous damages. Owning regulated infrastructure brings steady income, and it also brings tail risk that the headline P/E ratio doesn't capture.
What this means for long-term investors: a single week of relative performance is not a thesis. The real lesson is about what each kind of asset is priced to deliver. Berkshire's recent underperformance came when the market was paying up for AI growth, and its outperformance came when that premium shrank. For patient investors, the useful questions are structural. How much of a portfolio's return depends on one capital-spending cycle paying off? How much cash, and how many businesses that earn money in any market, sit underneath it? Owning some durable compounders and cash-generative businesses won't win every quarter. It does limit the damage when the market's favorite story stumbles. The next data point arrives in about a week, when Berkshire's 13F filing shows whether it changed its Alphabet position during the third quarter.