Which Artificial Intelligence (AI) Stocks Are Billionaires Buying the Most?

A review of Q3 2025 13F filings for 10 billionaire-run funds shows Alphabet and Nvidia were the most widely accumulated AI names, with half of the billionaires adding one or the other. Notable moves include Berkshire (Buffett) initiating a significant new position in Alphabet, Paul Tudor Jones increasing his Nvidia stake more than sevenfold, Millennium boosting Nvidia by 126%, Citadel adding 1.73 million Nvidia shares (a 21% increase), and Soros Fund Management more than tripling its Microsoft position; Broadcom, Meta and Microsoft also saw new or increased positions across several funds. The allocations highlight concentrated billionaire conviction in AI leaders—supporting continued investor interest in Google Cloud/TPUs and Nvidia GPUs—while not implying immediate changes to financials or near-term earnings guidance.

Analysis

Market structure: Billionaire buying concentrates capital into large-cap AI equities (NVDA, GOOG/GOOGL, MSFT, META, AVGO), widening bid-ask spreads favoring high-liquidity mega-caps and pressuring smaller AI names via fund flows. Nvidia (NVDA) and Alphabet (GOOG/GOOGL) are immediate winners — NVDA for GPU-driven AI compute demand and GOOG for cloud/TPU competition — while legacy ad-dependent peers (some META exposure) face higher execution risk as investors re-rate platform monetization vs. cloud/compute plays.

Risk assessment: Key tail risks include regulatory action on AI models/data (probability medium over 12–36 months), a 2026 chip slowdown from cyclical enterprise capex (low-to-medium), or a 20–30% correction if growth rerates. Short term (days–weeks) expect momentum-driven outperformance; medium (3–12 months) hinge on earnings beats in Google Cloud/NVDA datacenter; long term (2–5 years) depends on TPU adoption, Nvidia roadmap, and autonomous/quantum commercialization.

Trade implications: Favor size in NVDA and GOOG but stagger entries — 50% now, 50% on 8–12% pullback — and use protective options: buy 6–12 month NVDA call spreads (e.g., 1:1 3–6 month ITM/OTM) or 18-month GOOG LEAPs 10% OTM to capture structural upside while capping premium. Consider pairs: long GOOG vs short META to express cloud/AI infra over ad-monetization risk; sell covered calls on MSFT/META to monetize elevated implied volatility.

Contrarian angles: The market underestimates TPU-led displacement risk to Nvidia in specific AI workloads — a 3–5 year scenario where Google Cloud margins and vertical AI services compress third-party GPU demand. NVDA valuation is vulnerable to mean reversion; if AI capex normalizes, expect 30–50% downside from peak in a stress case. Conversely, Alphabet’s diversification (Waymo, quantum, XR) is underappreciated and offers asymmetric upside with lower execution multiples than priced.

Key Decisions for Investors

• Establish a 2–3% portfolio long position in NVDA now, add another 1–2% if price drops 8–12% within 3 months; hedge with 6–9 month 10% OTM put protection sized to 25–50% of the position cost to limit tail loss.
• Initiate a 1.5–2.5% position in GOOG/GOOGL via 18-month LEAP calls ~10% OTM (or equivalent shares if options illiquid), expecting cloud/TPU adoption to drive >30% upside over 24 months; average in on any 5% drawdowns.
• Enter a pair trade: long GOOG (1.5%) / short META (1.5%) to express infrastructure/AI compute over ad-monetization risk; rebalance after quarterly earnings or if the spread moves >15% against the position.
• Sell covered calls on existing MSFT or newly bought MSFT position to generate 3–6% annualized income (target 1–3 month calls, ~5% OTM) while maintaining core exposure to AI/cloud secular growth.
• Reduce non-mega-cap AI equity exposure by 20–40% and redeploy into NVDA/GOOG or cash if macro indicators (US ISM services <50 or 2y Treasury >4.5%) flip within 60 days, as liquidity and margin pressure will favor mega-cap winners.