The stock market just did something eerily similar to the dotcom bubble top in 2000

Only 20 S&P 500 members hit all-time highs as the index itself closed at a record, underscoring exceptionally narrow leadership concentrated in AI and semiconductor names. Bank of America’s Michael Hartnett warned the pattern echoes the dotcom bubble top and said central banks and rising rates could eventually end the speculative run. Breadth remains weak, with just about 55% of S&P 500 constituents above their 200-day moving average and advance-decline internals deteriorating since mid-April.

Analysis

This is a breadth problem before it is a headline-level market problem. When leadership compresses into a tiny cluster of AI/memory names, the index can keep grinding higher while the median stock quietly de-risks; that creates a fragile tape where passive inflows and systematic trend-following do most of the work until they don’t. The second-order effect is that portfolio managers who are underweight the winners but long the laggards get forced into a worse future entry point once breadth finally broadens or cracks.

The more important read-through is not just “AI is strong,” but that the trade is becoming an increasingly crowded capital-allocation event across the semiconductor supply chain. Memory, foundry, and accelerator names are now linked by a common narrative of AI capex durability; that raises the odds of synchronized disappointment if hyperscaler spending slows even modestly over the next 1-2 quarters. In that scenario, the first-order losers are the high-beta suppliers with the least pricing power, while the less crowded beneficiaries are quality defensives and duration assets that typically catch flows when breadth rolls over.

The risk window is asymmetric: near term, momentum can absolutely persist for days to weeks because narrow leadership is self-reinforcing and bearish breadth can stay bearish for longer than value investors expect. But over months, the setup is vulnerable to a rates-driven regime change or a simple pause in AI capex growth, either of which would expose how dependent the index has become on a small number of names. The contrarian miss in the market is that the “bubble” signal is not a timing tool; it is a positioning warning that upside is still possible, but the marginal dollar of return is increasingly coming with worse drawdown characteristics.

BofA’s defensive roadmap is directionally right, but I’d be more selective than a generic long-bonds/defensives call. If yields back up on sticky inflation or a hawkish repricing, long duration can get hit again even as breadth deteriorates, so the cleaner trade is to own sectors with low capex, low earnings dispersion, and limited AI dependence rather than making a blind duration bet. That favors utilities, staples, and certain healthcare/telecom names over high-duration software and the most crowded semiconductor beneficiaries.

Key Decisions for Investors

• Trim tactical longs in AMD and MU into strength over the next 1-3 weeks; keep only core exposure. Risk/reward is deteriorating because the same AI capex narrative that drove the move can unwind together, creating correlated downside if hyperscaler spending guidance softens.
• Initiate a relative-value pair: long XLU or XLP, short SMH on any further index strength. Hold 1-3 months. The trade expresses breadth deterioration without requiring an outright market short; downside on the short leg is capped if semis keep melting up, while defensives should benefit if rotation starts.
• Add BAC as a hedge-sensitive barbell only on a pullback, not chase it here. Banks can benefit if yield volatility stays elevated, but the position should be sized as a rates-volatility proxy rather than a pure growth bet; use a 6-12 week horizon.
• Buy 2-4 month put spreads on QQQ or SMH after a fresh momentum extension day. The market can stay narrow longer than expected, so define risk tightly; this is best as a convex hedge against a breadth-led air pocket rather than a standalone directional short.
• Watch for any 200-day breadth improvement as the trigger to cover defensive positioning. If participation expands materially over 2-3 weeks, the bearish breadth signal loses force and the market may reaccelerate beyond the current narrow leadership set.