AI Sector Shaken in a Single Week: Korea's Kospi Drops 10%, Investors Shift to Stock Picking
0xBroomberg
The AI trade just had its most volatile week yet — Korea's Kospi crashed twice, the Nasdaq 100 fell 3.3% in a single session, yet selling concentrated in mega-cap tech while other sectors drew inflows. This means → the AI thesis is intact, but the era of everything rallying together is over, and stock-picking is back.
What exactly happened this week?
Korea's Kospi fell 10% on Tuesday on AI chip-demand fears, then dropped another 5.8% on Friday after Microsoft and Apple price hikes stoked AI-inflation concerns.
The Nasdaq 100 lost 3.3% on Tuesday — its second-worst session of the year.
Micron briefly calmed chip-demand fears with a better-than-expected sales outlook, sparking a global tech bid, but the bounce did not hold.
How crowded are the positions?
Goldman Sachs prime-brokerage data show that as of June 18 — before the pullback began — AI and data-center long positions sat at a six-month high, but had not reached the extreme levels that signal a bull-market top.
This means → positioning is elevated but not at the "exit now" threshold; a systematic retreat has not started.
John Flood, head of Goldman's Americas equity execution services, said AI infrastructure stocks will drive roughly half of the S&P 500's EPS growth this year and called Q2 earnings the next key validation point.
Why was the volatility so extreme?
Goldman traders flagged thin liquidity near the top — in plain terms = at these elevated price levels, order books are sparse, so any shock triggers outsized price swings.
Volatility-control funds — strategies that automatically cut exposure when market swings rise — were forced sellers on down days, while leveraged ETFs held by Korean retail investors amplified the moves.
Hedge funds pushed their U.S. tech exposure back near a five-year high, yet their net long-short positioning in the Magnificent Seven — Apple, Microsoft, Nvidia and four other mega-cap tech names — fell to a one-year low. This reflects institutional money actively de-risking the most crowded names.
What is happening to the Magnificent Seven?
On Tuesday, even as the S&P 500 fell, the advance-decline ratio posted one of its strongest readings since 2023. This means → selling was concentrated in mega-cap tech; other sectors were actually absorbing capital.
The Magnificent Seven index has turned negative year-to-date. Alphabet is down 15% from its May peak and is losing key AI-model researchers after completing an $85 billion equity raise this month.
Memory-chip stocks are up more than 200% this year and data-center and power-equipment names have more than doubled, but the hyperscale cloud giants — Amazon's AWS, Microsoft's Azure and peers — have clearly stalled.
Where should investors look next?
Igor Pejic, author of *Blockchain Babel*, argues the AI industry is going through the value-migration that accompanies every technology revolution.
In plain terms = in the internet era, the big money was not made by Cisco, which built the routers, but by Amazon and Google, which ran on top of them. AI's growth momentum is shifting from chips and data centers toward vertical application companies — biotech, robotics and defense tech are leading examples.
This means → investors risk falling into a "value trap" by over-committing to the infrastructure layer. Finding the application-layer companies that actually monetize AI is the central stock-picking challenge of the next phase.
Content is for reference only, not financial advice.