Cathie Wood Goes AI Bargain Hunting: She Just Bought a Stock That Crashed 17% in 1 Trading Session and a Stock That's Dropped 50% From Its Peak.

Cathie Wood’s Ark Invest bought shares of Advanced Micro Devices and CoreWeave amid recent AI sector weakness, adding AMD across five Ark funds after the stock fell 17% on Feb. 4 despite an earnings beat; AMD guided to $9.8 billion in revenue for the quarter and now represents ~3.9% of Ark Innovation and trades near 29x forward earnings (down from >60x). CoreWeave, which debuted last March and is up ~80% since IPO but ~50% off its peak, was added to two Ark funds after reporting triple-digit revenue growth; its next quarterly report is scheduled for Feb. 26. The purchases signal Ark’s risk-on, long-term view on AI infrastructure names and may influence investor flows into AI chip and GPU-focused businesses despite valuation and spending sustainability risks.

Analysis

Market Structure: The immediate beneficiaries are GPU providers and GPU-capacity resellers — NVDA remains the pricing and architectural leader, AMD (AMD) is a close second for training/inference GPU TAM share, and CoreWeave (CRWV) captures variable-capex customers. Pricing power stays with constrained GPU supply and software lock-in; expect cloud/data-center customers (MSFT/GOOGL/AMZN) to pass cost into higher AI service fees, supporting gross margins for infrastructure owners over 12–36 months. Cross-asset: risk-on flows from renewed AI buying should push real yields up modestly (+10–30bp over weeks) and keep tech vols elevated; short-dated options on AMD/CRWV will remain rich around earnings (implied vol +20–50% vs 3-months prior).

Risk Assessment: Key tail risks are: renewed export controls on high-end GPUs (probability ~15–25% over 12 months), a macro shock that cuts enterprise AI spend (<=20% probability but high impact), and capital-starvation for hyperscaler-reseller models (CoreWeave burn forcing dilution). Near-term (days–weeks) volatility centers on earnings/GUIDANCE (CRWV 26-Feb, AMD next quarter); medium-term (3–12 months) hinges on supply (TSMC capacity) and pricing; long-term (1–3 years) is secular adoption if model-training spend grows into trillions. Hidden dependencies include Nvidia silicon roadmaps, foundry allocations, and cloud customers’ procurement cycles.

Trade Implications: Tactical long exposure to AMD at current ~29x forward earnings makes sense with 12–18 month horizon: initiate 2–3% portfolio long, scale to 4% on pullbacks >25% and set a 20% stop. CRWV is a higher-risk growth play: consider a 0.5–1% starter position and add only on Feb 26 beat + durable margin guide; hedge with short-dated puts if adding pre-earnings. Relative-value: pair long AMD / short INTC (1.5:1) to express GPU share gains vs legacy CPU weakness. Use call-spreads or 9–12 month LEAPs for directionally efficient exposure and sell near-term vols ahead of earnings to fund buys.

Contrarian Angles: The market may underprice capital-intensity: many GPU-capacity resellers will need recurring equity or debt raises, creating dilution risk (CRWV downside >40% if refinancing stalls). AMD’s 29x forward multiple already reflects a big reset from 60x; upside requires >15–20% revenue beat cadence next four quarters. Historical parallel: 2017–2019 GPU cycles showed rapid demand spikes then mid-cycle margin compressions; expect similar boom-bust amplitude. Hedge AI longs with 6–12 month out-of-the-money puts keyed to export-control headlines or a 25% drawdown trigger.

Key Decisions for Investors

• Establish a 2–3% portfolio long position in AMD (ticker: AMD) over the next 2 weeks, target 12–18 month hold; scale to 4% if shares drop another 20–25%; place a stop-loss at -20% from entry and reassess if guidance misses by >5% of revenue.
• Initiate a high-conviction, tactical 0.5–1% position in CoreWeave (CRWV) only after its 26-Feb earnings show revenue growth >40% QoQ and gross margin expansion or guidance that reduces capex intensity; pre-earnings exposure limited to selling a small number of covered calls or buying puts to cap downside.
• Implement a pair trade: long AMD (size X) and short Intel (INTC) (size 0.67X) to express GPU-led share gains vs legacy CPU stagnation; horizon 6–12 months, target relative outperformance of +15% and unwind if AMD underperforms guidance by >7% or INTC posts surprise AI roadmap wins.
• Use options to express asymmetric bets: buy AMD 9–12 month 25–35% OTM call spreads (funded by selling 1–2 week elevated-vol puts) and, around CRWV earnings, buy a short-dated straddle/strangle only if liquidity/pricing allows, otherwise wait for post-earnings volatility collapse to purchase directional LEAPs; cap options allocation to 1–2% of portfolio.