With Warren Buffett retired, Stanley Druckenmiller is now arguably one of the most closely watched money managers on Wall Street -- and his Duquesne Family Office just gave its followers plenty to chew on. The fund's latest quarterly filing with regulators, which captures its U.S. stock holdings as of March 31, revealed a clean break from one of the market's favorite names: Druckenmiller sold every last share of Alphabet (GOOG +4.96%)(GOOGL +4.79%), a stake he had aggressively built up just one quarter earlier. He also cut the bulk of his Amazon common stock position.
In their place, the billionaire rotated into the unglamorous hardware that the
artificial intelligence (AI) boom runs on. The wager seems to be that the next leg of AI spending will lean less on training enormous models and more on inference -- actually running them at scale -- which leans heavily on memory and custom chips. It's a bold move from one of the great track records in the business, so the natural question is whether everyday investors should do the same.
Image source: Getty Images.
The picks and shovels he bought
Druckenmiller disclosed new positions across a handful of AI hardware names that fall into two clear camps.
The first is memory and storage, where pricing is in the middle of a once-in-a-cycle surge as AI data centers absorb every bit of capacity the industry can produce. He opened positions in flash-memory specialist Sandisk (SNDK 1.98%), memory chipmaker Micron Technology (MU +0.90%), and hard-drive maker Seagate Technology (STX +7.63%).
These companies' growth has been staggering. Sandisk's fiscal third-quarter revenue (the period ended April 3) more than tripled from a year earlier to $5.95 billion, with its data center business alone up 233% from the prior quarter. Micron's most recent quarterly revenue nearly tripled as well, and management guided for current-quarter revenue to grow more than 200%. Seagate, meanwhile, grew revenue 44% last quarter while posting record margins.
These companies are also locking that demand in. Sandisk CEO David Goeckeler called the quarter "a fundamental inflection point" in the company's fiscal third-quarter earnings release, citing a shift toward higher-value data center customers backed by multiyear commitments. And Seagate says its highest-capacity drives are nearly spoken for through 2027.
The second camp is custom silicon. Here, Druckenmiller bought semiconductor giant Broadcom (AVGO +2.03%) and chip-design specialist Arm Holdings (ARM +2.79%). Broadcom designs the custom accelerators that big cloud companies use as alternatives to Nvidia chips, and that business is booming; AI revenue jumped 106% year over year to $8.4 billion last quarter, with more strong growth guided for the period it will report in early June.
Why following him may be risky -- and why Alphabet still looks attractive
There's a catch baked into every one of these filings: a six-week reporting lag. Druckenmiller's purchases reflect where the fund stood at the end of March, but the filing didn't surface until mid-May. In between, these stocks went vertical. Sandisk is up several thousand percent over the past year, while Micron has climbed more than 850% and Seagate has climbed about 600%. Anyone copying the trade now is paying far more than he did.
That gap matters because memory and storage are historically cyclical businesses. The forward price-to-earnings ratios on names like Sandisk and Micron look almost absurdly cheap -- in the single digits -- but that is often how deeply cyclical stocks appear at the top of a cycle, when current profits are peaking. Should supply catch up with demand, pricing and margins could reverse in a hurry.
It's also worth remembering how Druckenmiller tends to operate. He has a long history of ringing the register on winners early, sometimes too early -- he exited Nvidia in late 2024 and later called it a mistake -- and he has sounded wary about AI valuations for a while.
NASDAQ: GOOG
Key Data Points
Then there's the stock he left behind. Alphabet's business reported a stellar quarter since Druckenmiller unloaded the stock. First-quarter revenue rose 22% to $109.9 billion, its 11th straight quarter of double-digit growth, while Google Cloud revenue accelerated to 63% growth and its cloud backlog roughly doubled to more than $460 billion. Additionally, Alphabet's operating income rose 30%.
At a forward price-to-earnings ratio of about 27, Alphabet is no longer the bargain it was a year ago. Still, this valuation looks reasonable considering the search giant's business momentum.
So should you follow Druckenmiller out of the search giant and into AI hardware? I'm skeptical. His memory and custom-silicon bets may keep working, but they're cyclical, already up enormously, and being mirrored weeks after the fact. Alphabet, by contrast, still seems to offer durable double-digit growth at a fair price.
