Tech Stocks Rout $1.3T; AI Valuation Skepticism Hits Chip Stocks, Samsung/SK Hynix Down 12%

Global tech stocks experienced a sharp selloff on June 23–24, with Wall Street rotating out of AI darlings into defensive names as investors shift from rewarding AI spending to demanding evidence of outsized returns. Alphabet fell 5%, SpaceX plunged 16% (retreating from $200+ peaks in the weeks after its IPO), and NVIDIA dropped 3%. South Korea's tech-heavy Kospi index closed down 10%, with Samsung and SK Hynix each shedding 12% amid regulatory scrutiny on the semiconductor sector. The pan-European Stoxx 600 Technology index fell 3%, with chipmakers STMicroelectronics and ASML both down more than 7%.

The rout reflects Wall Street's recalibration on AI unit economics: although OpenAI's ChatGPT and Anthropic's Claude have splashed with consumers, only about 3% of Bank of America customers—mostly those with $125k+ annual income—pay for AI services. The Nasdaq had gained 26% from March 30 through the market's prior close, while the PHLX Semiconductor Index had advanced more than 100% over the same five-month period. Rising interest-rate expectations also pressured growth stocks: the Federal Reserve opened the door to rate hikes in 2026, and traders now price in nearly 90% odds of at least one increase by year-end, up from 57% odds a week prior.

Sentiment swung sharply: traders dumped highest-beta names like Strategy (MSTR), AppLovin, and Lumentum, then rotated into peanut butter and paint—Smucker jumped double digits, Home Depot and Sherwin-Williams led—along with real estate, staples, and utilities. Wedbush's Dan Ives called the pullback an opportunity for investors, though nervousness was amplified by Micron's looming earnings report.

For architects and investors: the tech rout signals a reset on AI capex multiples. Teams shipping AI products must now prove unit economics; chip allocations face near-term demand destruction unless cloud hyperscalers can demonstrate TCO improvements and pricing power. Watch for NVIDIA's next guidance, TSMC's fab utilization, and cloud-native inference cost curves to re-establish conviction.