Billionaire Stanley Druckenmiller has been bullish on AI since before ChatGPT launched. He is one of the most closely followed macro investors on Wall Street. He averaged about 30% a year at Duquesne Capital with no losing year, and his 1992 bet against the British pound made more than $1 billion.
Druckenmiller’s AI Stock Picks Take a Beating
The AI trade has taken a turn for the worse since the end of the second quarter. Nine of Stanley Druckenmiller’s 10 AI stock picks on this list have declined since June 30, with the stocks losing an average of 27.8% on an equal-weight basis through November 9, compared with a gain of about 0.8% for the SPDR S&P 500 ETF Trust (SPY). We have included each stock’s performance since June 30 in the list below.
But did Druckenmiller get out before the crash? We could find out in just a few days, when his family office is expected to disclose its third-quarter holdings. The filing will reveal whether he sold or reduced positions before the latest AI selloff, or held on to stocks that subsequently suffered steep losses.
To find out what Druckenmiller bought and sold, follow Stanley Druckenmiller’s page on our website and sign up for email notifications. You’ll receive an alert the next time we publish an article about his investments.
Why He Bought Nvidia Before the AI Boom
Druckenmiller told CNBC in May 2024 that his young partner urged him in 2022 to buy Nvidia. He bought it, and a month later ChatGPT launched, so he raised the position substantially. He cut the stake in late March 2024 after a huge run, and he said he is not Warren Buffett and does not hold stocks for 20 years. He called AI a little over-hyped in the short term and under-hyped in the long term.
Speaking of Buffett, we ranked the Berkshire Hathaway holdings with the most room to rise in our list of the 10 best Warren Buffett stocks with upside potential.
Is AI a Bubble? What Druckenmiller Has Said
Druckenmiller has called AI real from the start. In a June 2023 Bloomberg interview, he said AI is real, unlike crypto. He said real secular shifts do not move in 10 months, and that the internet bubble lasted about two and a half years. He wanted to own Nvidia for at least two to three years.
By May 2024, his tone had changed. He told CNBC that AI could be a little over-hyped now and under-hyped in the long term. He compared it to the internet in 1999, when someone who bought the Nasdaq watched it fall 80% before the internet’s payoff arrived. He said history could rhyme, and the big payoff from all the AI spending might be four to five years away.
What He Says About AI, Jobs and Inflation
In a Morgan Stanley interview recorded in January, Druckenmiller pointed out that people called every past technology revolution the end of jobs, all the way back to the horse and buggy. He gave radiologists as an example. Machines read prostate cancer pathology as well as or better than humans, but there are more radiologists today than ten years ago, because they now spend their time with patients. He also warned that AI could turn inflationary if governments answer job losses by printing money.
Methodology
Duquesne Family Office’s second-quarter 13F showed he is still buying AI stocks. In this list, we rank ten of those stocks by the value of Druckenmiller’s stake at the end of the second quarter, from lowest to highest. For each stock, we track its performance this year through November 9, measured from its January 2 close. We also calculated the returns since June 30th.
Why are we interested in the stocks that hedge funds pile into? The reason is simple: our research has shown that we can outperform the market by imitating the top stock picks of the best hedge funds. Insider Monkey’s quarterly newsletter strategy selects 14 small-cap and large-cap stocks every quarter and has returned 612% since May 2014, beating its benchmark by 375 percentage points (see more details here).
You can see the full list of AI stocks in Druckenmiller’s portfolio in our video by clicking the image below:
#10 Lam Research (NASDAQ:LRCX)
Druckenmiller Stake as of Q2: $18.9 Million
Year to Date Performance (through November 9): +19%
Return since June 30th: -48.9%
Lam Research makes the machines that chip makers use to build chips on silicon wafers, so every chip maker on this list depends on its equipment. Druckenmiller opened this position in the second quarter by buying 43,600 shares.
Lam gains from AI spending whichever chip maker wins, because it sells equipment to every major one. AI is driving a long memory boom, and chip makers need more equipment to meet that demand. Lam is at the center of it, with record sales and profit margins at all-time highs. Management expects the chip equipment market to keep growing into 2027, led by memory. Lam also plans to return most of its free cash flow to shareholders.
Supply is the main risk. Management sounded cautious about parts shortages and a lack of clean room space at chip makers, which can limit how fast customers add equipment. The stock fell after its latest report despite strong results. Sales to China are also expected to drop. The stock trades above its own historical valuation, which leaves little room for a slowdown in AI spending.
Lam Research’s profit grew strongly over the past year, but its cash from operations fell, which leaves less money for dividends and buybacks if the trend continues. Check the full valuation analysis here.
#9 Rambus (NASDAQ:RMBS)
Druckenmiller Stake as of Q2: $23.0 Million
Year to Date Performance (through November 9): -18%
Return since June 30th: -39.2%
Rambus makes memory interface chips. These chips sit next to the memory in AI servers and control how data moves between the memory and the processor. Druckenmiller opened this position in the second quarter by buying 173,000 shares.
Every AI server needs memory, so Rambus’s sales and profits keep growing. It has no long-term debt, a large cash pile and extra inventory to meet rising demand. Management expects data center demand to strengthen into 2027.
The stock fell after its latest report even though results and the forecast beat estimates, most likely because expectations were high and AI spending worries are hitting chip stocks. Bears point to slipping profit margins and rivals selling similar parts. Rambus also gains less from AI than memory makers such as Micron, since its chips are not scarce.
Rambus earns lower margins on its chips than on its licensing business, so faster chip growth can pull its overall profit margins down. Read the full breakdown here.
#8 Palo Alto Networks (NASDAQ:PANW)
Druckenmiller Stake as of Q2: $39.8 Million
Year to Date Performance (through November 9): +84%
Return since June 30th: -3%
Palo Alto Networks sells cybersecurity software and firewalls to companies. Companies now put AI agents inside their systems, and every agent adds another entry point that needs protection. Druckenmiller opened this position in the second quarter by buying 116,650 shares.
Bulls say AI is pushing security to the top of every company’s priority list, because each AI agent that connects to company systems is another entry point to protect. Identity security is becoming the control layer for those agents, and CyberArk gives Palo Alto a leading position there.
Bears acknowledge the critical role Palo Alto plays in AI security, but they point to valuation. The stock trades at 75 times expected earnings, against a sector median of 19.6 and its own five-year average of 59. Most of its recent revenue jump came from acquisitions, and organic growth is still about 15%. The deals also added about 19% more shares, so earnings growth is guided at just 9% in fiscal 2027.
Part of Palo Alto’s revenue growth came from acquisitions such as CyberArk, which also added shares, and its guidance points to earnings per share growth of only about 8% to 9% in fiscal 2027. See how the valuation holds up.
#7 AMD (NASDAQ:AMD)
Druckenmiller Stake as of Q2: $42.3 Million
Year to Date Performance (through November 9): +127%
Return since June 30th: -12.5%
AMD is Nvidia’s biggest rival in AI chips. Druckenmiller opened this position in the second quarter by buying 72,900 shares. The stake made up 0.81% of his portfolio. The stock is up 137.5% this year through November 9. It fell sharply over the past month and then rose on November 9 as Google Cloud picked AMD’s EPYC server chips for its new TPU system, and Oppenheimer upgraded the stock to outperform.
The case for AMD rests on one question: can it win AI customers from Nvidia? The latest evidence says yes. It has started shipping Helios, its full-rack AI system, to OpenAI, Meta, Microsoft, Oracle and Anthropic, and it has multi-year supply deals with OpenAI and Meta. Its server processors are also in demand, because AI agents need them to coordinate work. Management expects sales to grow more than 80% in 2027.
The stock already prices in much of that. It trades at 55.5 times expected earnings, against a sector median of about 19. Free cash flow fell last quarter as inventory piled up ahead of the Helios launch, and warrants issued to customers could dilute shareholders by up to about 16%. Nvidia’s software lead and custom AI chips from big customers are the other threats.
White Falcon Capital Management’s fourth-quarter 2025 investor letter laid out how AMD could keep climbing, after Jim Cramer praised CEO Lisa Su for moving the company aggressively into data centers. Read both views here.
#6 SanDisk (NASDAQ:SNDK)
Druckenmiller Stake as of Q2: $79.4 Million
Year to Date Performance (through November 9): +222%
Return since June 30th: -61%
SanDisk makes flash memory, the storage inside phones, laptops and AI data centers. Druckenmiller’s fund owns 34,900 shares, which made up 1.52% of his portfolio. He trimmed the stake in the second quarter by selling 3,255 shares.
Bears say the company’s growth will fade. Analysts forecast earnings growth of 221% in fiscal 2027 and only 27% in fiscal 2028. The stock trades at 3.7 times expected earnings, against a sector median of 19.6, which tells you investors do not trust these profits to last.
Bulls point to a real shortage. AI data centers need huge amounts of storage, and data center sales are SanDisk’s fastest-growing business. That demand drove the stock up this year.
Bears say shortages end. Chip makers in the US, South Korea and China are building new NAND factories, and China’s state-owned YMTC is already the world’s third-largest NAND maker by shipments. Data center spending could peak around 2028 as borrowing gets harder, just when that new supply arrives. In past NAND cycles, industry profits fell deep into losses after the boom. The stock is already about 60% below its June high.
SanDisk has long-term contracts with minimum prices that cover about half of next year’s output, but those contracts have never been tested in a memory downturn. Click to see the full bull and bear case.
#5 Seagate Technology (NASDAQ:STX)
Druckenmiller Stake as of Q2: $117.7 Million
Year to Date Performance (through November 9): +76%
Return since June 30th: -47.5%
Seagate makes the large hard drives that data centers use to store data. Druckenmiller more than doubled his stake in the second quarter by adding 71,300 shares. His fund now owns 122,000 shares, which made up 2.26% of his portfolio, up from 0.59% in the first quarter.
AI creates huge amounts of data, and Seagate’s largest drives are in short supply. The shortage lets Seagate raise prices, which has pushed its profit margins and cash flow to records. Its 2027 prices are already locked in, and Morgan Stanley thinks demand could grow even faster than the companies say.
The risk is 2028. Western Digital, Toshiba and parts suppliers are all adding capacity, just as Seagate sets its 2028 prices. Analysts still expect margins to keep rising. If AI spending slows, drive orders would feel it a few quarters later. The stock trades at 11.5 times expected earnings, a low multiple that shows investors expect profits to peak.
Seagate dropped more than 10% in a single day on a report that Toshiba plans to add hard drive supply, because more supply from a rival can weaken Seagate’s pricing power. See whether the dip is a buying chance.
#4 Alphabet (NASDAQ:GOOGL)
Druckenmiller Stake as of Q2: $120.2 Million
Year to Date Performance (through November 9): -9%
Return since June 30th: -19.3%
Druckenmiller sold his Alphabet stake in the first quarter and then changed his mind. His fund bought 336,300 shares in the second quarter, and the stake made up 2.31% of his portfolio. Alphabet is one of only two stocks on this list that are down this year.
The bull case is that Google Cloud has become a second growth engine. In the third quarter, Cloud revenue grew 103% from the same quarter a year ago, and CEO Sundar Pichai said demand for its AI infrastructure is huge. Nearly 100 customers use its TPUs, and Alphabet now sells them outside the company. Search revenue still grew 12% from a year earlier. Total revenue rose 25% and operating income rose 30% in the same period. Alphabet also owns about 5% of SpaceX, which went public in June. The stock is more than 30% below its May high, which gives buyers a lower entry point.
Alphabet’s free cash flow turned negative in the second quarter for the first time since the company went public, while Search revenue still grew 17% from a year earlier. Find out why investors still like it.
#3 Amazon (NASDAQ:AMZN)
Druckenmiller Stake as of Q2: $129.1 Million
Year to Date Performance (through November 9): +18%
Return since June 30th: 12%
Amazon Web Services rents out AI computing power to thousands of companies, and Amazon also builds its own AI chips. Druckenmiller’s fund added 495,800 shares in the second quarter and now owns 541,600 shares as of Q2. The stake made up 2.48% of his portfolio, up from 0.28% in the first quarter. His fund also held call options on Amazon shares.
Amazon could benefit even if the AI data center build-out slows after the midterm elections, Wall Street analyst Dan Ives said in a note to clients. He assumes about 10% of the 800 to 1,000 data centers planned in the US do not get built, and he tested a harsher case where 30% are cancelled over the next 18 months, according to the note. In either case, Amazon and other big cloud companies that already have data center capacity running become more valuable, because less new capacity competes with them, Ives wrote. He expects Big Tech spending growth to peak in 2027, earlier than his old forecast of 2028 or 2029, and he expects the big companies to raise about 25% less capital over the next year, Ives said. The AI boom still continues at a slower pace, which he called “hugely bullish” for Amazon, Microsoft, Alphabet and Meta Platforms.
L1 Capital expects Amazon Web Services spending to push Amazon’s free cash flow negative, but the fund says the market is too focused on near-term cash flow. Click to see why the fund stays optimistic.
#2 STMicroelectronics (NYSE:STM)
Druckenmiller Stake as of Q2: $232.4 Million
Year to Date Performance (through November 9): +62%
Return since June 30th: -40.4%
STMicroelectronics is a European company that makes chips for cars and industrial equipment. It now also sells chips to AI data centers, where they manage power and move data. Druckenmiller’s fund added 490,000 shares in the second quarter and now owns 3,102,880 shares.
STMicroelectronics bull case moves around a recovery play with an AI boost. Profits fell hard during the car and industrial slump, so the stock offered a chance to buy before earnings caught up. AI data center demand gives it a second growth driver. In its recently reported quarter, revenue rose 17.9% from the same quarter a year ago and beat Wall Street’s estimate. Sales in the segment that holds AI and satellite parts grew 46.2% from a year earlier. Management raised its 2026 data center sales outlook and expects 2027 to be much higher. It also expects factory changes to add more than 4 percentage points to gross margin by early 2028.
STMicroelectronics expects its AI data center sales to roughly double next year and is closing four older factories to lift its profit margins, though its power chip business still loses money. Check the risks and the upside here.
#1 Taiwan Semiconductor Manufacturing (NYSE:TSM)
Druckenmiller Stake as of Q2: $281.6 Million
Year to Date Performance (through November 9): +22%
Return since June 30th: -18.4%
TSMC makes chips for Nvidia, AMD and Apple, so much of the AI boom runs through its factories. Druckenmiller’s fund owns 589,680 shares, the largest stake on this list. It made up 5.40% of his portfolio, up from 4.96% in the first quarter. He added 94,400 shares in the second quarter, a 19.1% increase. The stock is up 22% this year through November 9.
Taiwan Semiconductor Manufacturing (TSM) remains well-positioned to benefit from long-term AI growth, regardless of which chip designer emerges as the industry leader. Its technological dominance in advanced semiconductor manufacturing, including its N2 process and CoWoS packaging technology, gives it significant pricing power and a competitive advantage that rivals may struggle to replicate. Demand for advanced chips continues to exceed supply, with production capacity largely booked through 2027. The company also benefits from strong profitability, software-like margins and substantial cash generation. Bulls say the recent semiconductor selloff could present an attractive entry point for long-term investors.
READ NEXT: 33 Stocks That Should Double in 3 Years and 15 Stocks That Will Make You Rich in 10 Years.
