Why healthcare is the 'ultimate hedge' against an AI correction, according to Kepler Cheuvreux

High-profile investor Michael Burry has amplified talk of an AI bubble by taking short positions in prominent AI-exposed names including Nvidia and Palantir, raising concerns about investor positioning in the sector. Kepler Cheuvreux analysts say healthcare is recovering from a recent 'perfect storm' and is increasingly benefiting from AI adoption that is driving efficiencies and innovation, according to Arnaud Girod.

Analysis

Market structure: Burry's public short of NVDA and PLTR amplifies downside risk for the most crowded AI longs while boosting demand for hedges and safe-haven sectors. NVDA retains structural pricing power for datacenter GPUs (high barriers to entry), but sentiment-driven flow can compress intraday liquidity and widen spreads; Palantir, with concentrated government/enterprise revenue, is more vulnerable to re-rating. The supply/demand signal is mixed — hardware demand remains strong long-term, but near-term inventory and hyperscaler capex cadence create a stop-start demand profile that raises volatility across semis and software.

Risk assessment: Tail risks include accelerated export controls on advanced GPUs, a governance or data-privacy shock to PLTR, or a sharp hyperscaler capex pullback — each could shave 20–40% off market caps in weeks. Immediate (days) risk is flow-induced volatility; short-term (1–3 months) risk centers on earnings/guidance; long-term (12+ months) favors secular AI adoption, especially in healthcare. Hidden dependencies: NVDA’s margins hinge on foundry/packaging continuity and hyperscaler multi-year commitments; PLTR depends on renewals and govt procurement cycles. Key catalysts: NVDA/PLTR quarterly guidance, US export policy updates (30–90 days), and major healthcare AI approvals.

Trade implications: Reduce concentrated NVDA exposure and buy time-protective insurance rather than outright exits; consider small asymmetric short on PLTR. Implement pair trades: long healthcare exposure (XLV) vs short PLTR to express rotation into durable AI beneficiaries. Use options to manage convexity — buy 60–90 day put spreads on NVDA to cap downside and buy outright puts on PLTR for directional short; sell 30-day call spreads on NVDA only if comfortable with assignment to fund hedges.

Contrarian angles: Consensus treats NVDA and PLTR as interchangeable 'AI plays' — that misses NVDA’s quasi-duopoly in accelerators and PLTR’s lumpy revenue sustainability issues. The market may be overpricing near-term destocking risk while underpricing healthcare’s multi-year AI productivity gains. History (2000–2002 tech unwinding) shows dominant infrastructure suppliers recover faster than hype software names; the unintended consequence of headline shorts is forced liquidity that can create buying opportunities in NVDA-sized moats.

Key Decisions for Investors

• If overweight NVDA, trim position size to a max 2–3% of portfolio within 5 trading days and purchase a 90-day put spread sized to protect ~50% notional (buy 7% OTM put, sell 15% OTM put) to limit downside through the next earnings (~60 days).
• Establish a tactical 1–2% net short exposure to PLTR funded by selling a non-core holding: buy 3‑month 15% OTM puts (target 30–40% downside within 6 months), set strict stop-loss at +25% premium paid and exit on revenue/renewal beat or renewed government contract visibility within 90 days.
• Rotate 3% of portfolio from AI-mega cap momentum into healthcare: buy XLV over next 2–6 weeks (target hold 6–12 months), take profits at +20% or re-assess if AI-driven margin expansion evidence appears in 2 consecutive quarters.
• Use options income to finance hedges: sell 30-day 5–10% OTM call spreads on NVDA sized to finance the put spread above (only if comfortable with potential assignment) and avoid naked short volatility into known catalysts; re-evaluate after next 4 earnings releases.
• Monitor three triggers in next 30–90 days before materially increasing directional exposure: NVDA revenue/guidance vs consensus, any US export-control announcements on GPUs, and PLTR renewal/win-rate disclosures; act (add/remove) if any trigger deviates >5–10% from expectations.