The catalyst today is not just that U.S. tech is red. It’s that the selling started rolling across time zones like a crack moving through the same sheet of ice.
By midday in the U.S., the Nasdaq Composite is trading at 25,843, down 1.2% from yesterday’s 26,167, while the Nasdaq 100 is off 2.1%. The S&P 500 is holding up better, down 0.9% at 7,407 versus yesterday’s 7,473, and the Dow is lower by just 0.2% at 51,603 versus 51,713 yesterday. That split matters. This is not a generalized growth scare; it is more concentrated in the long-duration, AI-heavy end of the equity market. The rise in the $VIX to 19.26 from 17.28 yesterday reinforces that this is a real risk-off move, not random tape noise.
The overseas setup makes the move more interesting. Europe’s STOXX 600 hit a more-than-one-week low as rate-hike fears and tech selling weighed on the region, with chip-linked names under pressure across the board, as reported here. In Asia, the damage was sharper: Japan’s Nikkei 225 fell 3.5% to 69,788, while a broader selloff in global chip stocks was tied to rate anxiety and concern that AI-linked valuations had run too far, too fast, as described here.
That cross-border pattern is the story. Semiconductor investing has become a global chain of confidence. Foundries, memory makers, equipment suppliers, packaging houses, networking vendors, and data-center landlords all trade on the same master assumption: AI demand will stay hot enough, long enough, to absorb aggressive spending. When the market starts questioning that assumption in Tokyo and Amsterdam, it rarely stops politely at the water’s edge.
The usual lazy explanation would be “higher-for-longer hurts tech.” That is not wrong, but it is incomplete. In fact, U.S. Treasury yields are softer today, not higher: the 10-year yield is around 4.48%, down from 4.51% yesterday, and the 5-year yield is about 4.25%, down from 4.29%. If lower yields are not cushioning semis, the issue is not discount rates alone. It is earnings durability at current multiples.
That distinction matters for names like NVDA, MU, ASML, TSM, KLAC and LRCX. The market has spent the last year treating the AI buildout as a near-linear demand story. Real businesses are never linear for long. Memory pricing cycles still cycle. Equipment orders still bunch up. Customers still digest inventory. Capex still competes with free cash flow, even when management teams dress it up in the language of inevitability.
That is why tomorrow’s MU report matters more than another intraday swing in a megacap favorite. Micron said it will report fiscal third-quarter results on June 24, in its earnings release notice, and investors will be watching whether high-bandwidth memory demand and broader DRAM pricing still support the valuation logic embedded across the AI supply chain. Micron’s prior fiscal second-quarter report showed revenue of $8.05 billion and diluted EPS of $1.41, both signs of a business recovering hard with AI demand in the background, as the company detailed here. The question now is not whether demand exists. It does. The question is whether demand is still surprising positively enough to justify everyone else’s expectations.
There is a second-order issue here too. A global semiconductor selloff tends to hit the most crowded part of the market first, but it does not always stay there. Today the equal-weight S&P 500 is down just 0.6%, better than the cap-weighted Nasdaq complex. If this remains a narrow unwind in expensive AI beneficiaries, broad equities can live with it. If it spreads into industrial demand assumptions, power equipment, and enterprise spending, then today’s move starts looking less like healthy skepticism and more like estimate risk.
What to watch: when MU reports, do investors hear confirmation that AI memory demand is still outrunning supply at attractive economics, or do they hear the first credible hint that the market’s favorite capex story is becoming just another cyclical business again?