The smart money isn't abandoning AI. They're just pricing in the odds it won't be enough.
The Summary
- Bank of America's Sebastian Raedler says investors should pivot to defensive sectors if the AI trade starts to unravel, signaling rising institutional concern about concentration risk
- The AI trade thesis remains intact, but the margin for error has shrunk dramatically as this week's volatility exposed how unforgiving the theme has become
- The shift marks a transition from "one thing" AI plays to a fragmented landscape where picking winners matters more than riding the wave
The Signal
The AI trade isn't dead. It's just not a layup anymore. For two years, you could throw money at anything with "AI" in the pitch deck and watch it climb. That era is over. The case for artificial intelligence as a transformative force hasn't changed, but the risk of getting the execution wrong has spiked.
This week's market action laid it bare. Sharp swings hit AI-adjacent stocks while leaving broader indices relatively stable. The message: investors still believe in the infrastructure buildout, the model improvements, the enterprise adoption curve. They just don't believe everyone building on that curve will make it.
"The risk of getting it wrong has risen considerably."
BofA's Raedler isn't calling for an AI collapse. He's calling for a rotation plan if the theme loses steam. That's the tell. When strategists at major banks start sketching exit routes, it means the easy consensus trade is fragmenting. Defensives, utilities, healthcare, the boring stuff that grinds higher when growth narratives stumble, those are the recommended hedges.
The shift reflects a maturation. Early-stage tech themes reward conviction and concentration. Late-stage themes reward selectivity and risk management. We're crossing that threshold. The "own the whole AI stack" strategy worked when capital was cheap and every layer looked like a winner. Now capital costs more, and investors are starting to ask harder questions:
- Which chipmakers actually have pricing power beyond this cycle?
- Which model builders have sustainable moats versus commoditized inference?
- Which enterprise software companies are genuine productivity multipliers versus feature bloat with a chatbot?
The Implication
If you're holding AI exposure, this isn't a sell signal. It's a sort signal. The companies building foundational infrastructure, the ones with recurring revenue tied to agent deployment, the platforms where customer lock-in compounds, those still work. The also-rans betting on narrative momentum without unit economics don't.
Watch the defensives. If utilities and healthcare start outperforming tech on a rolling three-month basis, that's your canary. It means institutional money is rotating out of growth and into stability, which historically precedes either a market correction or a prolonged chop. Either way, the days of rising AI tides lifting all boats are behind us.