Billionaire Stanley Druckenmiller Dumped Shares of Nvidia and Palantir and Is Loading Up on His New Favorite Artificial Intelligence (AI) Stock

Stanley Druckenmiller's Duquesne Family Office disclosed in its latest 13F filing a significant shift in its AI investment strategy, selling off its positions in Nvidia and Palantir, despite their substantial growth driven by the AI boom. The fund liquidated 9,500,750 Nvidia shares and 769,965 Palantir shares, potentially due to concerns over increasing competition in the GPU market and Palantir's high valuation. Concurrently, Druckenmiller increased his stake in Taiwan Semiconductor Manufacturing (TSMC) by 457%, making it a top-10 holding, betting on TSMC's critical role in chip fabrication for AI and its diversified revenue streams beyond AI, though tariff risks and valuation concerns remain.

Analysis

Stanley Druckenmiller's Duquesne Family Office, as per its Q1 2024 13F filing, has notably reconfigured its artificial intelligence (AI) investments, divesting entirely from high-growth AI stocks Nvidia (NVDA) and Palantir Technologies (PLTR), while substantially increasing its stake in Taiwan Semiconductor Manufacturing (TSM). The fund reportedly liquidated its entire holding of 9,500,750 Nvidia shares (adjusted for the June 2024 stock split) over a period culminating by the recent filing, despite Nvidia's 827% stock surge since early 2023 driven by its H100 and Blackwell GPUs dominating AI data centers. This sale may reflect concerns over rising competition from Nvidia's own major customers developing in-house AI GPUs, evidence of waning pricing power suggested by a declining gross margin, and the overarching risk of an AI bubble bursting, which would heavily impact Nvidia due to over 90% of its fiscal Q4 revenue stemming from its data center segment. Similarly, Duquesne shed its entire position in Palantir (769,965 shares), a company that has demonstrated over 25% sales growth and recurring profitability through its Gotham and Foundry platforms. The decision to exit Palantir likely centers on its exceptionally high valuation, with a trailing price-to-sales (P/S) ratio of 103, a level historically unsustainable and significantly above the typical peak P/S ratios of 30-40 for previous game-changing innovators. Concurrently, Druckenmiller acquired an additional 491,265 shares of TSMC, increasing the fund's stake by 457% and elevating TSMC to a top-10 holding. This move signals a strategic pivot towards a foundational AI enabler, as TSMC is a leading chip fabricator critical for AI chip production, currently expanding its CoWoS capacity from 35,000 monthly units in 2024 to an anticipated 135,000 by 2026 to meet demand from clients like Nvidia, AMD, and Broadcom. TSMC's diversified revenue streams, including manufacturing chips for smartphones (e.g., Apple's iPhone) and other non-AI applications, offer a potential buffer against a specific AI sector downturn. However, TSMC faces potential headwinds from unresolved tariff issues that could impact its operating margins, given most of its advanced chip production is in Taiwan, and its current valuation shows a P/S ratio over 10, a 16% premium to its five-year average, despite a reasonable forward P/E of 18.

Key Decisions for Investors

• Investors should critically assess their exposure to highly valued AI stocks such as Nvidia and Palantir, considering the potential for increased competition, margin pressures, and valuation normalization, as highlighted by Druckenmiller's divestment.
• Consider Taiwan Semiconductor Manufacturing (TSM) as a strategic investment in the broader semiconductor and AI ecosystem, given its critical role in the supply chain and diversified end-market exposure, while remaining vigilant about geopolitical tariff risks and its current valuation premium.
• It is prudent to monitor institutional investor filings (13Fs) for further insights into AI sector positioning and to closely track key industry indicators like GPU market dynamics, foundry capacity utilization, and potential signs of an AI investment cycle maturing.