Back to News
Market Impact: 0.55

Billionaire David Tepper Loaded Up on Nvidia Shares in 2025. Should You Follow Him?

NVDAAAPLGOOGLNFLXNDAQ
Artificial IntelligenceTechnology & InnovationCorporate Guidance & OutlookCorporate EarningsInvestor Sentiment & PositioningCompany FundamentalsMarket Technicals & Flows
Billionaire David Tepper Loaded Up on Nvidia Shares in 2025. Should You Follow Him?

Appaloosa Management’s David Tepper materially increased his Nvidia stake in 2025, growing the position from 300,000 shares at the end of Q1 to 1.9 million by end of Q3, making NVDA his fund’s fourth-largest holding. Nvidia CEO Jensen Huang said cloud GPUs were "sold out" in Q3 fiscal 2026 (ending Oct. 26), and the company projects global data-center capex rising to $3–$4 trillion annually by 2030 (versus roughly $600 billion for 2025), underpinning sustained demand for GPUs. Shares are roughly 10% below their all-time high and trading at ~24x next-year earnings, which the author frames as a buying opportunity ahead of expected monster growth in 2026 and beyond.

Analysis

Market structure: Tepper's >6x build to 1.9M NVDA shares and NVDA trading ~24x forward implies concentration of capital into a company that sits atop a multi-year data‑center capex cycle (company estimate $3–4T/year by 2030 vs ~$600B in 2025). Direct winners: NVDA, TSM/ASML/Samsung (foundry, lithography, HBM suppliers), cloud hyperscalers; losers: commoditized CPU vendors (INTC), smaller GPU hopefuls and any vendor unable to secure HBM/TSMC slots. Pricing power should stay strong near term because unit supply (silicon + HBM + ASML capacity) is the choke point, not demand. Risk assessment: Tail risks include tightened US/China export controls, a hyperscaler demand pullback (elasticity shock), or a supply‑side overbuild (memory/fab capacity) — each could cut revenue growth >30% year-on-year in a stress case. Immediate (days) sensitivities are to NVDA quarterly guidance and export headlines; short-term (weeks/months) to TSMC/TSMC capacity announcements and HBM spot prices; long-term (2026–2029) to hyperscaler vertical integration and capex pace. Hidden dependencies: power/data‑center build timelines, HBM supply and ASML delivery slots; catalysts that accelerate trend include multi‑year GPU purchase commitments from top 3 hyperscalers or a major foundry capacity expansion announcement. Trade implications: Core tactical: establish a 2–3% long NVDA position now, plan to add to 4–6% on a >15% pullback within 3 months; trim if position appreciates >40% or forward PE >30x. Pair trade: long NVDA 3% vs short INTC 1.5% to express clear margin/AI monetization dispersion, re-evaluate quarterly. Options: buy a Jan‑2027 NVDA 12–18 month call spread (e.g., 400/700) sized to 1% portfolio risk to capture upside while limiting theta bleed. Rotate 1–2% into ASML/TSM as supply‑chain hedges with 12–36 month horizon. Contrarian angles: Consensus underestimates the fragility of supply and overestimates perpetual market share — hyperscalers (GOOGL, AMZN) could accelerate in‑house silicon (historical parallel: memory cycles where capex led to price crashes). The market may be underpricing regulatory/export tail risk and overpricing multiple expansion: set explicit cut triggers (e.g., forward PE >30x or a 10-day realized IV spike >50% alongside guidance miss) to avoid momentum trap. Unintended consequence: massive NVDA skews sector flows, pressuring valuations across tech and compressing yields in long‑duration bonds if AI capex expectations reprice higher.