If the AI Bubble Is About to Burst, 2 Things Warren Buffett Would Recommend Investors Do Right Now

The article warns that AI stocks have made the S&P 500 and Nasdaq-100 extremely top-heavy and tightly correlated, raising downside risk if hyperscaler AI spending slows. Valuation metrics are flagged as frothy: the Buffett indicator is over 235% (vs. a “reasonable” 75%-90% and “overvalued” above 120%) and the CAPE ratio is above 42 vs a ~17.4 long-term average. The suggested mitigation is Buffett-style positioning—keep cash on the sidelines and favor companies with durable moats (citing Alphabet as a potentially resilient AI exposure if a pullback occurs).

Analysis

The real signal here is not “AI is expensive” — it’s that index risk is now functionally a capped call on a small set of capex-heavy names. That creates a reflexive loop: if hyperscaler spending slows even modestly, semis, networking, power, and data-center beneficiaries can de-rate together, while passive vehicles mechanically amplify the drawdown because the same few names dominate the benchmark. In that setup, the first-order loser is NVDA, but the second-order loser is market breadth itself.

GOOG looks better than the other AI complex because it monetizes AI through a durable distribution layer rather than just capex conversion. BRK.B is the cleaner “crash preparedness” expression: not because it is a hedge, but because it offers dry powder when forced sellers create dislocations. The key distinction is that the article’s valuation metrics are timing-poor; they justify caution, not an all-out short unless earnings revisions start rolling over.

Catalysts over the next 1-3 months are hyperscaler capex commentary and whether AI revenue is showing up in operating leverage rather than just spend. The thesis is falsified if the next earnings cycle confirms accelerating inference demand, stable cloud margins, and continued capex expansion; in that case, the concentration is a feature, not a bug. Over 6-18 months, the risk is less a single crash than a sharper-than-normal unwind in crowded mega-cap exposure once breadth deteriorates and passive flows reverse.

Key Decisions for Investors

• Initiate a small QQQ or SPY put spread into the next 1-3 earnings cycles as a portfolio hedge; target ~1-2% premium spent, since the goal is convexity against a crowdedness unwind rather than a standalone crash call.
• Pair trade: long GOOG / short NVDA for 1-3 months as a relative-quality expression; GOOG has better downside support from distribution and balance-sheet optionality, while NVDA is most exposed if AI capex growth rate decelerates.
• Accumulate BRK.B only on a broad-market pullback of 5-8% or if megacap breadth weakens materially; this is a cash-optionality trade, not a momentum trade.
• Set a watch item on hyperscaler capex guidance and AI-related capex-to-revenue conversion; if growth rates inflect down for two consecutive quarters, reduce NVDA/QQQ exposure aggressively.
• If you are already underweight mega-cap tech, do nothing on this article alone; it is a positioning warning, not a fresh catalyst.