AI Crowded Trade Unwind: Funds Rotate from Semiconductors to Value Stocks
Miles Bennett
UBS's AI-beneficiary basket underperformed its AI-disrupted basket by 42 percentage points this month, a record gap; this is rotation, not panic — and Big Tech earnings will decide when it stops.
What does a 42-point gap actually tell us?
UBS's basket of AI-beneficiary stocks (including semis) trailed its basket of AI-disrupted stocks by 42 percentage points this month — a record.
This means → capital that piled into AI over recent months was too concentrated; once conviction cracked, the unwind hit just as hard.
In plain terms = everyone crowded onto the same boat. Now it has capsized, and the stocks nobody wanted — the "old economy" names — are the ones running.
How far has the momentum trade unwound — and is it over?
Momentum trading — a strategy that rides the strongest-performing stocks — has given back all gains since February, while volatility keeps climbing toward pandemic-era levels.
Goldman Sachs equity-execution head Brian Garrett says the "momentum unraveling is in its back half," but notes realized volatility remains "strikingly elevated."
This means → the most violent phase may be past, but the tail end will still be rough; Goldman sees room to add AI exposure from here.
Why is this rotation, not a crash?
Bank of America derivatives strategists note the tech pullback "once again looks like rotation, not indiscriminate selling" — healthcare and financials clearly outperformed.
Even within the "Magnificent Seven," dispersion is stark: Apple — lower AI exposure, more value character — rallied, while momentum favorites Alphabet and Nvidia fell.
This reflects a market that has not rejected AI itself, but is punishing the most crowded positions. Value is the clear beneficiary.
Is the ASML sell-off an overreaction?
ASML sold off after reports that a Chinese state-linked firm has begun manufacturing immersion DUV lithography equipment.
Industry analyst Masahiro Wakasugi calls this likely an overreaction: DUV — deep-ultraviolet lithography, used for less-advanced chip nodes — is far less sophisticated than ASML's EUV tools — extreme-ultraviolet lithography, higher precision, currently irreplaceable — essential for cutting-edge AI chips.
He estimates China still needs roughly seven to ten years to fully close the gap with ASML; the threat of a 20% hit to ASML's sales and profit "still looks a long way off."
Why hasn't a record earnings season rescued the market?
Of S&P 500 companies that have reported so far, 85% beat estimates — the highest rate in five years of Bloomberg Intelligence tracking — yet the market has not rallied.
Capital.com senior analyst Daniela Hathorn says "investors are growing pickier: strong revenue growth is no longer enough unless accompanied by evidence that heavy spending is converting into sustainable profitability."
In plain terms = the market no longer asks "how much did you earn?" It asks "is the money you're pouring into AI actually coming back?"
What comes next?
Bank of America recommends a call-spread strategy on SMH (the semiconductor ETF) to position for a rebound — "renting" upside via options rather than owning it outright, capping downside risk.
This means → even Wall Street isn't buying the dip with conviction; instead it's using options — a "small-bet, defined-risk" tool — to test the water.
Apple and Amazon report next. Whether the remaining mega-caps can deliver convincing evidence that AI capex is paying off will determine if this rotation stops here.
Content is for reference only, not financial advice.