Stanley Druckenmiller does not stay in a trade long once it has done its work. He manages the Duquesne Family Office, has one of the most respected macro investing records in modern history, and built that record largely by rotating capital early into the next structural shift before the crowd arrives.
His latest disclosed trades show exactly that playbook in action. Druckenmiller has exited SanDisk entirely and moved his capital into an energy company that most investors in the AI space have barely noticed yet.
How much SanDisk Druckenmiller sold, and why the timing matters
Druckenmiller sold 166,235 shares of SanDisk after holding the position for just one quarter. During the period he owned SanDisk, the stock surged more than 400%, according to Motley Fool. That is not the kind of return that most investors walk away from. But Druckenmiller is not most investors.
SanDisk was spun off from Western Digital in February 2025, making it an independent company at precisely the moment AI-driven demand for high-speed storage and NAND flash memory was accelerating fastest. Hyperscalers racing to build training clusters created an almost ideal tailwind for SanDisk’s core flash business, and the stock reflected that reality emphatically.
Fund manager buys and sells
- Cathie Wood buys $2.5 million of tumbling megacap stock
- Warren Buffett dumped 77% of Amazon to buy surging media stock
- Cathie Wood buys $11 million of tumbling megacap tech stock
But a stock that has surged 400% in a single quarter has already priced in a great deal of optimism. Memory chips are cyclical. The same pricing environment that creates enormous earnings upside for SanDisk can compress quickly if supply catches up to demand or if the rate of AI infrastructure buildout slows.
Druckenmiller appears to have decided that the easy phase of the SanDisk trade was behind him.
Druckenmiller buys Bloom Energy instead
After exiting SanDisk, Druckenmiller initiated a new position in Bloom Energy. Bloom Energy builds solid-oxide fuel cells that convert natural gas into electricity. The stock has risen more than 800% since its 2018 IPO.
The investment thesis is straightforward. AI data centers are consuming electricity at a scale that the existing power grid was not built to handle. The five largest AI hyperscalers have shared plans to make up to $720 billion in capital expenditures in 2026 alone, much of it allocated to new data center capacity, Motley Fool noted.
Many of those projects face permitting delays and grid interconnection queues that can stall construction for years.
Bloom Energy’s fuel cells can be installed behind the meter, bypassing the grid entirely and providing firm, dispatchable power that complements renewable sources. That characteristic makes Bloom Energy directly relevant to the power problem that is becoming the central bottleneck in AI infrastructure expansion.
The company has already secured deals with Oracle, CoreWeave, and Equinix to test or deploy Bloom Energy systems at new data center campuses, Motley Fool confirmed. On the policy side, the Trump administration’s focus on energy abundance and faster permitting creates a favorable environment for natural gas solutions that can scale quickly.
The Bloom Energy buy says something specific about AI
Druckenmiller’s move is not a retreat from the AI trade. It is a reallocation within it. He sold SanDisk, a memory company that benefits from AI-driven storage demand, and bought Bloom Energy, a power company that benefits from AI-driven electricity demand.
The theme is identical, but the bottleneck he is betting on has shifted.
Related: Billionaire Druckenmiller buys $152 million in megacap tech stocks
That distinction matters for investors trying to understand where the next leg of AI infrastructure spending flows. Memory and compute were the dominant constraint in the early phases of the buildout.
Power is increasingly becoming the binding constraint now, and the companies that solve the power problem are beginning to attract exactly the kind of institutional attention that previously went to chipmakers.
Druckenmiller’s history suggests he spotted this transition early. He exited SanDisk after a 400% gain when other investors were still adding to the position, and he rotated into Bloom Energy before its profile in AI infrastructure conversations had fully arrived in mainstream coverage.
That sequencing is consistent with how he has operated across decades of successful macro trades.
Key figures from Druckenmiller’s disclosed trades:
- SanDisk shares sold: 166,235, after one quarter of holding
- SanDisk return during Druckenmiller’s holding period: More than 400%
- SanDisk IPO context: Spun off from Western Digital in February 2025
- Bloom Energy stock performance since 2018 IPO: Up more than 800%Â
- Bloom Energy technology: Solid-oxide fuel cells converting natural gas to electricity, deployable behind the meter
- Bloom Energy data center customers: Oracle, CoreWeave, and Equinix
- AI hyperscaler capex plans for 2026: Up to $720 billion
- Druckenmiller background: Former manager of George Soros’s Quantum Fund, now runs Duquesne Family Office
Source: Motley Fool
Nagle/Getty Images
What Druckenmiller’s exit means for SanDisk investors
Druckenmiller’s exit is not a verdict on SanDisk’s fundamentals. Morgan Stanley analyst Joseph Moore recently raised his SanDisk price target to $1,100 from $690, projecting 51% revenue growth in 2026 and seeing meaningful upside from the ongoing NAND pricing cycle. The company reports Q3 2026 earnings on April 30, and expectations are high.
But Druckenmiller’s sale after a 400% gain in one quarter is a reminder that even strong businesses with powerful tailwinds can become expensive relative to the opportunity that remains. Memory chips are cyclical. SanDisk’s earnings power is real, but the stock’s performance has already reflected a great deal of optimism about the duration of the current cycle.
Investors still holding SanDisk should not take Druckenmiller’s exit as a signal to sell. They should take it as a prompt to ask whether their own thesis for holding the stock still holds at current prices, and whether the risk-reward balance looks as attractive today as it did when the position was initiated.
What this means for Bloom Energy investors
Bloom Energy is not a widely followed name in AI infrastructure conversations. That may be exactly why Druckenmiller finds it interesting now. A company that solves a real problem before the market assigns it full credit is the kind of setup that has driven some of his most successful trades.
The power constraint in AI infrastructure is real and growing. Utility-scale grid connections involve multi-year permitting timelines. Data center developers who cannot secure reliable power cannot build.
Bloom Energy’s ability to deliver firm, modular power without grid interconnection gives it a structural advantage in exactly the environment AI infrastructure expansion is creating.
Whether Bloom Energy becomes the next major AI infrastructure trade depends on execution. The company has to keep winning customers, scaling production, and demonstrating its technology performs at the power densities modern data centers require.
Druckenmiller’s position is a signal that he believes it can. It is not a guarantee.
Related: Billionaire Stanley Druckenmiller’s buys point to tech stock shift