A sell-off like Friday's is ‘rarely the top’ of a rally, says Wharton’s Jeremy Siegel

Tech stocks sold off sharply as the Nasdaq fell 4.7% for the week, its worst weekly decline since April 4, 2025, while the VanEck Semiconductor ETF dropped nearly 5% for the week and 9.2% on Friday, its worst day since Jan. 27, 2025. Jeremy Siegel characterized the move as a common pullback after a parabolic rise rather than the start of a prolonged correction, though he warned that chip valuations need earnings to stay elevated. The article frames the decline as sentiment- and momentum-driven, with AI-related chips still a major year-to-date outperformer.

Analysis

This reads less like a macro regime change and more like a forced de-grossing event in a crowded factor basket. The first-order damage is in semis, but the second-order winner is likely quality growth outside the AI capex complex: if momentum funds are reducing exposure mechanically, capital should rotate toward beneficiaries with cleaner earnings durability and less dependency on a single narrative. That argues for relative strength in index-heavy platforms and data/market infrastructure versus the high-beta hardware chain.

The key risk is not the one-day drawdown itself, but whether it causes dealer hedging and CTA trend signals to flip from support to supply over the next 1-3 weeks. If that happens, semis can underperform fundamentals for longer than valuation models justify, because positioning unwinds faster than estimates reset. The critical tell will be whether the group can reclaim prior highs quickly; failure to do so would convert a normal shakeout into a path-dependent multiple compression.

The contrarian view is that the market may be underpricing the duration of AI earnings power, but overpricing its smoothness. The more important question is not whether AI spending continues, but whether the incremental dollar of capex still clears the cost of capital after the easy productivity gains are captured. That makes the near-term setup attractive for tactical longs in the strongest balance sheets, but hazardous for names where the market has already capitalized several years of perfect execution.

For the listed names, WisdomTree is largely a sentiment proxy and should benefit indirectly if the dip is interpreted as healthy rather than terminal, while Nasdaq’s franchise is less about immediate earnings leverage and more about volatility in trading/IPO activity if risk appetite stays elevated. Broadcom is the clearest pressure point because expectations around AI monetization are now more vulnerable to any hint of slower accelerator demand; that creates a potentially better entry only after forced selling finishes, not during it.

Key Decisions for Investors

• Tactically short SOXX or SMH for 3-10 trading days on failed rebound attempts; use tight risk if the ETF reclaims the prior breakout level, because the trade is a positioning unwind rather than a structural short.
• Pair long NDAQ / short SOXX for 1-2 months to express ‘risk-on survives, hardware cools’; the trade should work if market activity remains healthy while semi multiples compress.
• Sell cash-secured puts on AVGO only after 2-3 sessions of stabilization; preferred tenor 30-60 days, targeting a lower entry with positive theta while avoiding catching the falling knife.
• For longer-term accounts, buy the highest-quality AI infrastructure names on weakness only if breadth broadens; otherwise wait for a higher low, since parabolic leaders often need 1-2 failed retests before durable support forms.