Sorry, Dr. Burry. I'd Rather Own Nvidia at a Premium Than Adobe at a Discount

The article argues Michael Burry’s bearish put position on SOXX expiring in January could be early, but warns semiconductor shorts remain risky amid continued AI-driven strength and elevated volatility. It favors Nvidia as a steadier AI beneficiary at 32.2x trailing P/E, while viewing Adobe as a value contrarian play at 8.0x forward P/E that may still face a long road. Overall, the piece is a bearish-to-cautious take on semis and a mildly constructive view on select AI winners.

Analysis

The market is still treating semis as a one-way momentum factor, but the more interesting setup is not valuation — it’s inventory elasticity. If AI capex is even modestly pulled forward now, the system can overshoot on the way in and on the way out, creating a classic bullwhip over the next 2-4 quarters: foundry utilization stays high, then enterprise digestion forces order cuts that hit the group with a lag. That makes the downside path for hardware more violent than the average multiple compression story, but also means timing matters more than conviction.

NVDA remains the cleaner relative long because it is the only asset here with pricing power, ecosystem lock-in, and enough demand visibility to absorb a cyclical air pocket. The second-order winner is not just the chip supplier set, but any beneficiary of AI infrastructure spend that is less exposed to end-demand sentiment; if GPU demand stays tight, weaker software vendors face a narrative squeeze as investors rotate toward tangible capex beneficiaries. ADBE, by contrast, is vulnerable to being re-rated on growth duration rather than earnings power: even if fundamentals are stable, the market is likely to pay a lower multiple until it sees evidence that AI is additive rather than substitutionary.

The contrarian miss is that bearish semiconductor positioning can become self-defeating in the near term: any shallow pullback in a tape dominated by buybacks, passive flows, and retail dip-buying is likely to get crushed. But if the AI trade slows, the correction will likely be abrupt rather than linear, because positioning is crowded and implied volatility is still cheap relative to the potential for estimate revisions. That argues for using options and pair structures rather than outright shorts, with a medium-term horizon into the next earnings cycle where guidance, not headlines, will decide whether this is a pause or a regime change.

Key Decisions for Investors

• Own NVDA vs. SOXX as the cleaner expression of AI capex durability: long NVDA / short SOXX into the next 6-12 weeks, with the view that firm-level winners hold up better than the basket if the trade de-risks.
• Buy downside convexity in SOXX through Jan expiry puts or put spreads on rallies; target a 2:1 to 3:1 payoff if the semi complex reprices on any guide-down or inventory digestion headline over the next 3-6 months.
• Fade ADBE on strength with a medium-dated call spread overwrite or a small short against a basket of better-duration software names; the risk/reward improves if the market continues to punish software exposed to AI substitution risk.
• If NVDA sells off 10%+ on no fundamental change, use that as the entry for a tactical long: the stock still has the best support from demand visibility and should outperform on any recovery in risk appetite.
• Avoid naked short semis here unless you can tolerate a 20-30% squeeze; the cleaner expression is long-vol or relative-value, since momentum and passive flows can overpower fundamentals for multiple months.