Buffett Just Made Alphabet a Top-3 Holding — What the Q2 Filings Reveal About the Smart Money
The Q2 13F filings are in, and they tell one coherent story: the great investors are still all-in on AI — they've just stopped paying any price for it.
Every quarter, institutions managing above the reporting threshold must disclose their U.S. equity holdings within 45 days. The snapshot is dated — these are June 30 positions — and it can omit foreign-listed shares and distort option values. But read for the why rather than the what, this batch is unusually eloquent.
Berkshire: A Conviction Bet, and a Rescue Option
The headline is Berkshire's Alphabet purchase — roughly $17 billion added in the quarter, making it a top-three holding faster than any position since the Apple build. The logic tracks Buffett's oldest yardstick, return on invested capital, and the interesting part is that Alphabet's ROIC has been crushed lower by AI capex. Buying anyway is a statement: the mega-cap spending spree isn't waste, it's survival spending — the price of defending a franchise through a technological inflection. And the search franchise is holding. After an initial scare in which query growth stalled, AI-integrated search has pushed usage to record levels. There's also a quieter angle: a buyer of Berkshire's size doesn't just predict survivors — its capital can help make one, as it did for Goldman in 2008.
The second-biggest Berkshire story is Delta. Airlines were long a textbook bad business — returns below the cost of capital for decades. Delta's pivot changed the math: premium-cabin revenue has overtaken the main cabin for the first time, and the co-branded card and loyalty franchise now generates fast-growing, recurring, capital-light income. Berkshire buying an airline in a high-fuel-cost year is not nostalgia; it's a bet that this is no longer an airline-shaped business. Smaller adds in homebuilders and retail round out the filing.
The Pattern Across the Other Filers
| Investor | Bought | Sold / trimmed | The read |
|---|---|---|---|
| Berkshire | Alphabet (~$17B), Delta, homebuilders | — | Franchise survivors, transformed business models |
| Druckenmiller | Alphabet, Amazon, CDW, TSMC, STMicro | Broadcom, SanDisk (partial) | Keep AI, swap rich U.S. chips for cheaper ones |
| Hohn (TCI) | Martin Marietta, Vulcan, Alphabet, S&P Global | — | Moats only — including literal quarries |
| Fisher | GE Aerospace, GE Vernova, Cisco, Palo Alto | Chevron, ConocoPhillips, CNQ | AI infrastructure in; energy profits taken |
| Coatue | Micron, Intel | Lam Research, Applied Materials/ASML | Forward-cheap memory in, rich equipment out |
Three threads run through the table. First, nobody abandoned AI — every filer added exposure somewhere. Second, the selling was concentrated in the U.S. semiconductor names that had run hardest into the June peak of the chip index; the discipline was valuation on trailing numbers, on the logic that forward estimates minted during a mania are themselves part of the mania. Third, the buying migrated toward what you might call AI's byproducts: CDW, the IT reseller positioned to build AI systems for thousands of small firms that can't do it themselves; and Hohn's aggregates producers, whose quarries sit within trucking distance of the large majority of U.S. data-center construction — a moat issued by geology rather than by engineers. When direct AI beta gets expensive, the smart money buys the picks, shovels, and gravel pits.
What It Means for You
The lesson isn't any single ticker. It's that the most experienced capital in the world is simultaneously long the AI buildout and unwilling to overpay for its most crowded expression. If your AI exposure is concentrated in the handful of chip names everyone already owns, this filing season is a prompt to ask Buffett's question — which franchises survive the wave? — and Hohn's — where is the moat competition can't cross? — before the next leg of the cycle answers them for you.
Backtest momentum, market-cap, low-volatility, value and quality strategies across the whole S&P 500 — and see each one's weekly returns — in the free Quant Lab.
Open the Quant Lab →
Disclosure: Educational content only, published August 2026. This is not investment advice or a recommendation to buy or sell any security. Position sizes and figures are drawn from public quarterly filings and reporting and may be incomplete or superseded. Verify against the filings themselves before acting.