Griffin isn’t pulling out of AI, but he had good reasons to take some profits and rebalance his bets on the trend.
Ken Griffin is not the typical Wall Street billionaire. As founder and CEO of Citadel, he leads one of the most sophisticated and consistently profitable hedge funds in history. He has more than earned his reputation as a rare mix of macro strategist and quantitative mastermind.
Citadel’s second-quarter 13F form, filed in August, revealed a few notable moves: The hedge fund reduced its position in the data mining specialist Palantir Technologies (Fut -5.39%) by 48% – selling approximately 640,000 shares. That left Citadel with a stake that is worth about $130 million today. At the same time, the company strengthened its position in the semiconductor giant Nvidia (NVDA -4.84%) by a whopping 414% – adding more than 6.5 million shares and bringing its stake in it to approximately $1.5 billion at current share prices.
These moves speak volumes about Griffin’s changing view of the artificial intelligence (AI) landscape. Below, I’ll outline what might have influenced Citadel’s decision-making, and what investors can conclude from this reshuffling.
Why Griffin Cut Palantir: A Rational Rebalancing of Hedge Funds
At first glance, Citadel’s decision to divest nearly half of its stake in Palantir seems bearish. But in the world of hedge funds, selling doesn’t always indicate lost conviction; it is often about managing risks.
Palantir’s meteoric rise — shares are up more than 2,000% in the past three years — has made even its most sincere believers aware that the stock is perfectly priced. Palantir trades at a price-to-sales ratio of 135 and is flirting with valuations reminiscent of the dot-com era.
For disciplined managers like Griffin, emotion can’t undermine basic math. Hedge funds thrive by constantly reallocating their capital, which includes trimming positions in their winners once they’ve gone too far, too fast. Interestingly, famous investors like Stanley Druckenmiller and Cathie Wood have previously played the same hand with Palantir positions.
In general, taking some chips off a high-flying stock is simply about risk-adjusted returns. Palantir’s long-term fundamentals remain impressive, but hedge funds can’t afford to be sentimental. Locking in certain profits gives them more financial flexibility to invest in opportunities that offer a better balance between upside potential and appreciation.
Image source: Getty Images.
Why Citadel Continues to Buy Nvidia: The Backbone of the AI Infrastructure
In my opinion, Citadel’s growing position in Nvidia reflects Griffin’s belief that the next decade of computing will be defined not by those who write algorithms, but by those who control the infrastructure on which they run.
In recent years, Nvidia has become the world’s leading supplier of accelerated computer hardware. GPUs now power everything from large language models (LLMs) and autonomous systems to next-generation humanoid robots.
But Nvidia’s dominance is not just about hardware, but also about the ecosystem. The company’s CUDA software platform has become a lock on itself, locking developers into a computing standard that few rivals can match.
Citadel’s accumulation of Nvidia shares reflects several strengthening trends:
- Hyperscaler releases: Cloud infrastructure giants love it Microsoft, AmazonAnd Alphabet collectively invest hundreds of billions of dollars annually to expand data center capacity.
- Strategic integrations: Nvidia’s partnerships with OpenAI, IntelAnd Oracle are only beginning to realize their potential.
- Relentless innovation: The company’s upcoming GPU architectures – Blackwell Ultra and Rubin – expand its technological lead and will secure its growth trajectory.
Although Nvidia also trades at a premium valuation, its structural growth story remains unparalleled. As companies and governments race to build an AI infrastructure, Nvidia stands as the most reliable – and perhaps indispensable – foundation for the AI era.
A masterclass in portfolio evolution
Griffin’s moves indicate a calculated rotation within the AI megatrend – not a retreat from it. By cutting its stake in Palantir and doubling its bet on Nvidia, Citadel is effectively betting on where the next wave of outsized profits in AI will originate.
The message for investors is clear: don’t chase hype stories and don’t follow momentum. Instead, try to anticipate capital flows. Griffin doesn’t abandon AI; he’s rebalancing his bets on the trend toward the pick-and-shovel plays that support it.
My take is that Griffin thinks Palantir’s story is maturing, while Nvidia’s machine is still ready to add power.
Adam Spatacco holds positions at Alphabet, Amazon, Microsoft, Nvidia and Palantir Technologies. The Motley Fool holds positions in and recommends Alphabet, Amazon, Intel, Microsoft, Nvidia and Palantir Technologies. The Motley Fool recommends the following options: a long call in January 2026 at $395 on Microsoft, a short call in January 2026 at $405 on Microsoft, and a short call in November 2025 at $21 on Intel. The Motley Fool has a disclosure policy.