Billionaire Ken Griffin Sells Amazon Stock and Buys an AI Stock Up 1,030% Since 2024 (Hint: Not Nvidia)

In Q3 Citadel’s Ken Griffin trimmed Amazon by selling 1.6 million shares and initiated a 388,000-share position in Palantir, a stock up ~1,030% since January 2024. Amazon reported revenue up 13% to $180 billion, operating income up 23% to $21.7 billion and a 60 bp expansion in operating margin (ex one‑offs), with Street EPS growth ~18% CAGR and a current ~33x earnings valuation. Palantir posted 63% revenue growth to $1.1 billion and $0.21 non‑GAAP EPS in Q3 but trades at ~119x sales (versus 18x in Jan 2024), implying extreme multiple expansion versus fundamentals and raising valuation risk despite strong AI demand.

Analysis

Market structure: AI demand is bifurcating winners — infrastructure/cloud providers (AMZN, MSFT, NVDA) gain sticky, high-margin revenue while specialist apps (PLTR) get speculative multiple expansion from narrative-driven flows. Amazon’s 13% revenue growth, expanding operating margin and 33x forward EPS imply durable cash-flow optionality; Palantir’s 119x sales implies extreme duration risk — price is driven by multiple expansion, not fundamentals (revenue <2x vs. 11x price since Jan). Cross-asset: a tech-led rally compresses IG spreads and raises growth equity duration; a Palantir collapse would jump equity volatility and widen high-yield spreads briefly, pressuring levered quant funds and elevating FX flows into USD safe havens.

Risk assessment: Tail risks include a rapid multiple compression on PLTR (>50% within 1–3 months), US government contract cuts affecting Palantir revenue concentration, and regulatory scrutiny on data/AI that could hit both revenues and margins over 6–24 months. Immediate (days) risk is momentum reversal; short-term (weeks–months) risk is mean reversion of sentiment-driven names; long-term (years) centers on sustained product differentiation and unit economics. Hidden dependencies: PLTR relies on public sector wallet and ops integration; AMZN depends on continued AWS pricing leverage and enterprise AI adoption.

Trade implications: Favor overweight in AMZN (core cloud + ad exposure) and NVDA (AI infra) while de-risking or shorting sentiment-driven AI apps like PLTR. Use relative trades: long AMZN vs short PLTR to capture spread if PLTR multiple contracts. Options: buy 6–12 month AMZN LEAPS (e.g., Jan 2026/27 $150–$200 calls) sized 2–4% portfolio, and purchase 1–3 month PLTR put spreads (e.g., 10–25% OTM) to limit capital and capture a volatility decline.

Contrarian angles: Consensus underweights the fragility of PLTR’s multiple — but overestimates AMZN’s vulnerability; market may continue to bid PLTR higher on momentum and index flows, creating short-squeeze risk in next 2–6 weeks. Historical parallels: rapid multiple reratings (e.g., early Nvidia/NFLX runs) can persist before mean reversion; trade sizing must respect that. Unintended consequence: aggressive shorting of PLTR could be painful if retail and quant momentum funds sustain flows; prefer option-defined risk or pair trades to mitigate crowd squeezes.

Key Decisions for Investors

• Establish a 2–4% portfolio long position in AMZN within 0–6 weeks using 9–12 month LEAPS (Jan 2026 $150–$200 calls or equivalent) to capture AWS/advertising upside; hedge by selling 1–2% covered calls if held >9 months to monetize premium.
• Initiate a tactical bearish position on PLTR sized 1–2% of portfolio using 1–3 month put spreads (buy 20% OTM put, sell 10% OTM put) to profit from a volatility and multiple contraction; close if PLTR declines >40% or if implied vol drops below 80% of recent 30‑day average.
• Execute a relative-value pair trade: long AMZN (2%) vs short PLTR (2%) — target spread tightening of 30–50% in 3–9 months; unwind if AMZN underperforms PLTR by >15% on fundamental miss.
• Reduce generic ‘speculative AI app’ exposure (ex: non‑profit cash-flow names trading >30x sales) by 25% within 30 days and redeploy into AI infrastructure (NVDA, MSFT, AMZN) to lower portfolio duration risk and sensitivity to multiple compression.
• Monitor three catalysts over next 30–90 days before expanding positions: PLTR quarterly bookings and government contract renewals, AMZN AWS margins and Rufus monetization cadence, and 10‑year US Treasury yield moves >50bp which would materially repricing high‑growth multiples.