Billionaire Investor Offloads GPU Stocks, Makes Bold Bet on Semiconductor Alternative

Recent Form 13F filings shed light on a major strategic pivot in billionaire Philippe Laffont’s investment portfolio. The latest quarterly filing from Coatue Management reveals that Laffont has decided to sell significant GPU-heavy positions while simultaneously building a new core holding. This move represents a notable shift in how elite investors view the artificial intelligence landscape and where true value lies in the AI infrastructure boom.

The Strategic Shift: Why Laffont Chose to Sell GPU-Dominated Stocks

Wall Street closely monitors Form 13F filings—the quarterly snapshots of what sophisticated investors are buying and selling. Laffont’s filing, submitted in late February 2026, showed a decisive move to sell 667,405 shares of Nvidia and 253,768 shares of Meta Platforms during the fourth quarter. Yet this wasn’t a one-off decision. Over the past three years, Laffont has been methodically reducing exposure to these GPU-centric plays. His Nvidia stake has plummeted by 82% (approximately 40.6 million split-adjusted shares), while his Meta position has contracted by 53% (roughly 4.3 million shares) since March 2023.

The rationale behind selling these GPU stocks likely stems from multiple considerations. While both Nvidia and Meta possess formidable competitive advantages—Nvidia’s superior graphics processing units command the AI infrastructure market, and Meta’s social media ecosystem remains unmatched—their share prices have already delivered extraordinary returns. Nvidia shares have surged roughly 1,200% since 2023, while Meta has climbed approximately 445%. For a seasoned investor like Laffont, cashing in after such gains represents disciplined profit-taking.

There’s also speculation that Laffont harbors concerns about an AI valuation bubble. Historical precedent suggests that transformative technology trends often experience early-stage corrections when investors overestimate adoption timelines and optimization potential. Although AI infrastructure demand remains robust, realizing meaningful bottom-line improvements could take years.

The New Top Holding: Taiwan Semiconductor Manufacturing Emerges as the Winner

Rather than simply selling positions to raise cash, Laffont redirected capital strategically. Taiwan Semiconductor Manufacturing Company (TSMC) has now become his fund’s leading AI stock and overall top holding. During the fourth quarter, Laffont purchased approximately 556,988 additional shares, vaulting TSMC into the No. 1 position within his $40 billion portfolio.

This move makes considerable sense when examining TSMC’s role in the AI revolution. The company operates the world’s most advanced chip fabrication facilities, and it has been dramatically ramping monthly production capacity to meet insatiable demand for high-bandwidth memory integrated with cutting-edge GPUs. As long as GPU demand continues outpacing available supply, TSMC’s order backlog and pricing leverage should remain formidable.

Beyond AI: Why TSMC Represents Diversified Value

The true appeal of TSMC extends well beyond its AI-related revenue streams. Unlike Nvidia—which is heavily concentrated in GPU sales—TSMC functions as a crucial provider of wireless chips for next-generation smartphones, advanced semiconductors for Internet of Things applications, and automotive-grade processors. While these segments lack the explosive growth trajectory of AI operations, they collectively establish a stable foundation and deliver consistent cash generation.

This diversification may have been precisely what attracted Laffont. By transitioning from GPU-pure-play stocks to a more balanced semiconductor manufacturer, he reduces concentration risk while maintaining robust exposure to artificial intelligence themes.

The Valuation Equation: Is TSMC’s Premium Justified?

Laffont appears drawn to TSMC’s current valuation as well. Trading at a forward price-to-earnings ratio of 21, the company appears reasonably valued should it meet or exceed consensus expectations for 31% revenue growth in 2026 and 24% expansion in 2027. For a company commanding such technological superiority and operating leverage, this valuation multiple offers meaningful appeal compared to GPU-pure-play alternatives that have already appreciated substantially.

The decision to sell GPU-concentrated holdings and reposition into TSMC underscores how professional investors constantly reassess positioning. What appeared attractive at previous valuations may no longer offer compelling risk-reward dynamics, while overlooked alternatives can suddenly emerge as superior choices when examined through a fresh analytical lens.