Nearly 40% of Billionaire Bill Ackman's Hedge Fund Is Invested in 3 Monster AI Stocks

Pershing Square had over 39% of its portfolio in three AI-focused mega-cap tech names as of Dec. 31, 2025, owning $2.2B of Amazon, $1.9B combined of Alphabet (both classes), and $1.8B of Meta. Ackman entered these positions at attractive forward P/Es (Alphabet Q1 2023 at 16; Amazon Apr 2025 at 25; Meta Q4 2025 at 20) and cites strong AI-driven demand, large 2026 capex plans (Alphabet $175–185B; Amazon $200B; Meta $115–135B) and a combined $331B cash balance as justification for concentrated bets.

Analysis

Primary second-order winners are the upstream compute and infrastructure suppliers that scale with hyperscaler AI cycles: GPUs, high-bandwidth memory, power & cooling OEMs, and datacenter interconnect vendors. Expect lumpy, multi-quarter demand shocks to these suppliers as large model training windows concentrate capacity consumption; that creates predictable windows for price elasticity and inventory buildup that public markets often misprice.

A key risk is return-on-capital compression: incremental AI-driven revenue growth need not translate to proportionate free cash flow if pricing for compute, data center real estate, or ad inventory is competed away. Regulatory and ad-market shocks can materialize on quarters’ notice and will hit multiples faster than operating leverage rebuilds — downside scenarios are asymmetric in the 6–18 month window.

Trade construction should prioritize convexity: buy-leverage in the form of multi-quarter LEAPS or call spreads into pullbacks, and use short-dated hedges to protect against near-term sentiment shocks tied to guidance. Where the market prices consensus execution as binary, construct pairs to isolate idiosyncratic execution risk (e.g., hardware-maker vs. cloud operator) rather than directional macro exposure.

Contrarian angle: the market is implicitly pricing perpetual winner-takes-most monetization; it underweights the probability that hyperscalers will mutually commoditize premium LLM inference pricing to stimulate usage, compressing margins industry-wide. That makes hedges and relative-value shorts of lagging incumbents a higher-expected-value defensive allocation than outright overweights into stretched hardware multiple without protection.

Key Decisions for Investors

• Long NVDA / Short INTC pair — equal-dollar exposure, 6–18 month horizon. Rationale: capture asymmetric upside from GPU-led training demand vs Intel’s slower accelerator cadence. Risk/reward: target 40–80% relative upside on NVDA vs 25–30% downside in a semiconductor-cycle drawdown; stop-loss 20% absolute on NVDA leg.
• Buy GOOGL Jan-2027 LEAPS (or a 12–18 month call spread) sized 2–4% of portfolio, financed with 3–6 month covered calls at 10–15% OTM. Rationale: long-duration convexity to multi-year AI monetization; target 35–60% upside, max loss = premium paid (0–100% of option premium).
• Accumulation tranche for AMZN via buy-and-hold with a protective put calendar: buy shares or Jan-2028 LEAPS (3–5% portfolio), buy 3–9 month put spreads to limit near-term downside. Rationale: asymmetric upside from platform & cloud synergy; expected payoff 30–50% over 12–36 months versus capped downside (~10–20% with hedges).
• Tail hedge: buy META 6–9 month put spreads (tight strikes) sized to cover 1–2% portfolio exposure. Rationale: protect against capex/execution disappointment and ad softness that could remove the short-term valuation premium. Risk/reward: cost ~1–3% of portfolio for protection that pays ~6–10x on a severe drawdown within the hedge window.