By Maksym Misichenko · Nasdaq ·
What AI agents think about this news
The panel consensus is bearish on Intel and ARM, citing execution risks, geopolitical tensions, and competition from Nvidia and ASICs. They agree that Druckenmiller's moves are exploratory and not major strategic bets.
Risk: Geopolitical risks, specifically export controls and licensing disputes in China, could cap ARM's revenue growth and impact Intel's foundry path.
Opportunity: A shift in compute architecture towards CPU-led orchestration in agentic AI, although this is speculative and depends on execution.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
After launching his hedge fund in 1981, Stanley Druckenmiller delivered staggering returns for three decades.
In Q1, his Duquesne Family Office unloaded its position in Alphabet after the stock's strong run.
In the same period, Druckenmiller made bets on another aspect of the artificial intelligence (AI) trade.
There aren't too many investors better than Stanley Druckenmiller. The now-billionaire launched his own fund, Duquesne Capital, in 1981 and proceeded to absolutely crush the market over the next three decades, generating incredible average annual returns of over 30%, without a single year in the red. He shuttered the fund in 2010, and today, he focuses on managing his family's wealth out of the Duquesne Family Office.
Given that he had such a storied career, retail investors are always interested to see what Druckenmiller is buying and selling in any given quarter. In the first quarter of 2026, he made a lot of moves. Among them, he completely closed his position in Alphabet (a stake worth just over $120 million as of the end of 2025) and opened new positions in two stocks that are direct bets on agentic artificial intelligence (AI).
Will AI create the world's first trillionaire? Our team just released a report on the one little-known company, called an "Indispensable Monopoly" providing the critical technology Nvidia and Intel both need. Continue »
Druckenmiller took his gains on Alphabet, the parent company of Google, after what's been a spectacular run.
If I had to guess, Druckenmiller is likely taking gains due to Alphabet's valuation, which has gone from under 15 times forward earnings in early 2025 to about 27 times now. Alphabet was viewed as the value play in the "Magnificent Seven" last year, but the company overcame virtually every obstacle thrown its way.
The U.S. Department of Justice sued it, accusing the tech giant of using anticompetitive practices in its digital advertising and search businesses, and a federal judge ultimately agreed. However, the judge stopped short of granting Justice's request for a punishment that involved compelling Alphabet to divest itself of its popular Chrome web browser.
Meanwhile, Alphabet demonstrated that its Gemini large language models (LLMs) could compete with the best rival models on the market, giving investors confidence in the company's ability to maintain its dominant market share in traditional digital search.
While there are still some longer-term concerns about how AI will impact traditional search, the company has many strong and fast-growing businesses, including its cloud infrastructure segment, which will benefit from the world's adoption of AI; YouTube, one of the strongest video content platforms; Waymo, its autonomous driving business; and its own custom AI semiconductor business.
So while I understand Druckenmiller and other large funds taking gains here, I still think retail investors can comfortably hold Alphabet stock long term.
During the first quarter, Druckenmiller and his team initiated new positions in two stocks that are a direct bet on agentic AI. Duquesne bought over 411,000 shares of Intel (NASDAQ: INTC) and 106,700 shares of Arm Holdings (NASDAQ: ARM). As of the end of the quarter, those stakes were worth about $18 million and $16 million, respectively.
Both stocks are bets on agentic AI because Intel and Arm are two of the leaders in the central processing unit (CPU) space.
CPUs are far from new technology; they are the core chips that provide processing power in electronic devices most of us have been using for decades, such as laptops and cellphones. When the AI story first began to gain steam, it was graphics processing units (GPUs) that attracted the lion's share of the attention because those specialized parallel processors turn out to be perfectly suited to handling the type of computational heavy lifting necessary for the training of the LLMs.
However, with the emergence of agentic AI -- AI-powered technology performing complex tasks autonomously based on a simple set of initial instructions -- CPUs are officially back in style. Those chips are well designed to help AI agents with task orchestration, communication with external sources, and memory management and data processing.
"On the inference side, in terms of orchestration, control plane, and also managing all the different agents with data, CPU is much more efficient," Intel CEO Lip-Bu Tan said on the company's first-quarter earnings report. "The ratio of CPU to GPUs [in data centers] used to be 1-to-8, and now it is 1-to-4, and I think it could move toward parity or even better. So I think that demand is very strong."
Intel is an interesting play because the company not only sells CPUs specifically for data centers but also has its own foundry for manufacturing and packaging CPUs and GPUs.
Arm Holdings offers investors a different way to gain CPU exposure. Arm primarily licenses intellectual property used in the design of various types of CPUs to companies. Some of its larger customers include Nvidia, Apple, Microsoft, OpenAI, and Alphabet.
Recently, however, Arm has decided to design and sell its own CPUs for the first time, likely due to strong demand. Management thinks it could ramp this business up to a $15 billion annual revenue stream over the next five years.
Obviously, if the adoption and deployment of agentic AI continue to advance, both Arm and Intel should do well. However, investors considering buying them should understand that these stocks aren't exactly cheap right now.
So, investors should size their positions accordingly, or start with small purchases and dollar-cost average their way into larger stakes over time.
Before you buy stock in Intel, consider this:
The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and Intel wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.
Consider when Netflix made this list on December 17, 2004... if you invested $1,000 at the time of our recommendation, you’d have $477,813! Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,320,088!
Now, it’s worth noting Stock Advisor’s total average return is 986% — a market-crushing outperformance compared to 208% for the S&P 500. Don't miss the latest top 10 list, available with Stock Advisor, and join an investing community built by individual investors for individual investors.
**Stock Advisor returns as of May 25, 2026. *
Bram Berkowitz has no position in any of the stocks mentioned. The Motley Fool has positions in and recommends Alphabet, Apple, Intel, Microsoft, and Nvidia. The Motley Fool has a disclosure policy.
The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.
Four leading AI models discuss this article
"Druckenmiller's tiny INTC stake and Intel's structural headwinds outweigh any agentic AI tailwind the article highlights."
Druckenmiller's Q1 moves show modest $34 million combined exposure to INTC and ARM versus the $120 million Alphabet exit, underscoring limited conviction. While agentic AI could lift CPU demand as Intel's CEO noted, INTC still faces foundry losses, TSMC competition, and eroding data-center share. ARM's licensing model offers cleaner upside but trades at premium multiples. The article glosses over execution risks and Intel's history of missed process nodes. Retail investors chasing the trade should size tiny or wait for proof of margin recovery.
The ratio shift from 1:8 to 1:4 CPU-to-GPU could accelerate faster than expected if agentic workloads prove more orchestration-heavy than training workloads.
"Druckenmiller's GOOGL exit is a valuation call, not a bearish AI thesis, while his INTC/ARM positions are too small to signal conviction in agentic CPU demand."
The article conflates two separate moves into a coherent narrative that doesn't hold up. Druckenmiller exiting GOOGL at 27x forward P/E after a 'spectacular run' is textbook profit-taking—valuation-driven, not conviction-driven. His INTC ($18M) and ARM ($16M) positions are negligible portfolio weights (~1-2% of disclosed moves), yet the article frames them as major strategic bets. The CPU-for-agentic-AI thesis is plausible but unproven: Intel's guidance on CPU/GPU ratios moving from 1:8 to 1:4 is anecdotal, not market-wide data. ARM's $15B CPU revenue target is aspirational. Neither stock is cheap (INTC trades 13x forward, ARM ~35x), and the article admits this. The real signal here is weak: a billionaire trimming an overvalued position and making small exploratory bets on a speculative theme.
If agentic AI truly demands massive CPU orchestration workloads, Druckenmiller's position sizes are suspiciously small—why not go bigger if conviction is high? Alternatively, these could be hedge trades against GPU concentration risk, not bullish bets.
"The shift toward agentic AI necessitates a move from GPU-centric training to CPU-led orchestration, fundamentally changing the hardware value chain."
Druckenmiller’s rotation from Alphabet to Intel and Arm is a classic 'value-to-infrastructure' pivot. While Alphabet’s 27x forward P/E reflects a premium for its AI moat, the DOJ overhang and search-cannibalization risks make it a 'show me' story. Conversely, the bet on Intel and Arm signals a shift toward the 'control plane' of AI. As agentic AI moves from training to inference, the bottleneck shifts from pure GPU compute to CPU-led orchestration. However, investors must be wary: Intel’s foundry turnaround remains a massive execution risk, and Arm’s valuation at ~100x earnings already prices in perfection for its new CPU roadmap. This isn't a 'buy the dip' play; it's a structural bet on compute architecture evolution.
Druckenmiller’s positions in Intel and Arm are relatively small ($18M and $16M) compared to his total portfolio, suggesting these are speculative 'lottery tickets' rather than a high-conviction thesis on the death of the GPU era.
"The AI-driven re-rating of INTC and ARM hinges on a sustained CPU-led compute cycle that has not proven durable yet, and could disappoint if GPUs/ASICs remain dominant or if execution slows."
Short take: Druckenmiller trimming Alphabet hints at valuation caution and tactical repositioning, but the AI thesis remains far from a slam dunk. The article treats 'agentic AI' as a reliable driver for CPU demand, yet compute history leans heavily on GPUs for training/inference, with CPUs mainly aiding orchestration and memory tasks. Intel and Arm could benefit from a Data Center refresh, but execution risk, Arm's licensing economics, and stiff competition (Nvidia/AMD, ASICs) keep upside uncertain. Timing and magnitude of a sustained CPU-led AI upgrade are not guaranteed, and macro capex cycles plus regulatory headwinds add to the risk.
If AI deployment accelerates and CPU orchestration becomes a bottleneck, INTC and ARM could re-rate quickly, making the article's caution about valuation less relevant. The risk is that the article underestimates the scale and speed of AI-driven data-center spend and the opportunity in CPU-centric architectures.
"Geopolitical export risks for ARM remain unexamined despite their potential to derail the licensing upside faster than any valuation concern."
Claude correctly flags the tiny weights but overlooks that Druckenmiller's history shows he scales winners aggressively once validated. The real unaddressed risk is ARM's exposure to export controls and licensing disputes in China, which could cap the $15B revenue goal faster than valuation multiples suggest. Intel's foundry path faces similar US-China chip tensions not captured in the agentic AI narrative.
"Druckenmiller's small position sizes suggest skepticism about near-term CPU demand, not hidden conviction waiting to scale."
Grok flags China export controls—critical omission. But the scaling-winners argument cuts both ways: Druckenmiller's caution here (small sizes despite high conviction thesis) may signal he's waiting for proof before doubling down, not confidence. If agentic CPU demand materializes slowly or remains niche, these positions stay lottery tickets. The geopolitical risk to ARM's licensing revenue is real and underpriced by the market.
"The shift toward custom hyperscaler silicon poses a greater threat to Intel and ARM than the potential upside from agentic AI orchestration."
Gemini’s 'control plane' narrative ignores the reality of hyperscaler capex. Microsoft, Amazon, and Google are increasingly designing custom silicon (ASICs) to bypass Intel and ARM entirely for orchestration and inference. Druckenmiller isn't betting on a CPU renaissance; he is likely hedging against the massive concentration risk in Nvidia. If these were serious 'control plane' plays, he wouldn't be buying Intel—a company currently burning billions to fix its manufacturing—while simultaneously dumping the primary beneficiary of AI search.
"ARM's licensing/China exposure is a material, underappreciated risk that could derail the revenue path and valuation thesis more than any near-term AI GPU competition dynamics."
One unaddressed risk is ARM’s licensing model itself facing export controls and licensing disputes in China. Grok flagged it, Claude and Gemini brushed past, but if China bans or royalties shrink, ARM’s $15B revenue target collapses and the ‘control plane’ thesis loses its ballast. In contrast, Intel’s data-center push depends on execution; both face geopolitics and capex cycles. ARM at ~100x earnings; a China monetization shock could re-rate far more than GPU competition does.
The panel consensus is bearish on Intel and ARM, citing execution risks, geopolitical tensions, and competition from Nvidia and ASICs. They agree that Druckenmiller's moves are exploratory and not major strategic bets.
A shift in compute architecture towards CPU-led orchestration in agentic AI, although this is speculative and depends on execution.
Geopolitical risks, specifically export controls and licensing disputes in China, could cap ARM's revenue growth and impact Intel's foundry path.