|
The Fire Alarm Was the Jobs Report. The Exit Door Was Semiconductors.
On Monday morning, June 2, the S&P 500 closed above 7,600 for the first time. All three major indexes printed fresh records. The AI trade was running uninterrupted, with Marvell up 33% in a single session on commentary from Nvidia’s CEO.
By Friday afternoon, the Nasdaq had given back 5.1% from that peak. The Philadelphia Semiconductor Index had shed 10.3% in 48 hours. And the VIX, which opened the week below 16, closed at 21.51.
The easy explanation is Broadcom. The company reported fiscal Q2 revenue of $22.19 billion and non-GAAP earnings of $2.44 per share, both above consensus. But its Q3 AI chip sales guidance of $16 billion missed the $17.2 billion estimate, and it declined to raise its full-year AI semiconductor forecast. Shares fell 12% on Thursday and another 8% on Friday.
The real accelerant came at 8:30 a.m. Friday. The Bureau of Labor Statistics reported 172,000 jobs added in May. The Dow Jones consensus was 80,000 to 85,000. FactSet had it at 105,000. Goldman Sachs, which tends to run high, came in at 120,000. Nobody had 172,000. Leisure and hospitality alone added 70,000, more than the entire headline consensus. Local government added 55,000. And the BLS revised March and April higher by a combined 93,000 jobs.
Average hourly earnings rose 0.3% month-over-month. Not alarming. Unemployment held at 4.3%. Not alarming either. But those are backward-looking measures. The bond market priced the forward: the 10-year jumped to 4.54%, and the 2-year hit 4.17%, its highest since February 2025.
Read the sequence, not the headline. The chip trade was already wobbling on Thursday. Broadcom gave investors a reason to question near-term AI earnings growth. Then Friday’s payroll number gave the bond market a reason to reprice the rate path entirely. Two separate catalysts, one compounding outcome. The weakest hands in the most crowded trade had two consecutive reasons to leave.
The base case here is not that AI demand is broken. Broadcom’s revenue still grew. Nvidia’s addressable market, per its own CFO at the Bank of America conference last week, could double with its next-generation chips. The base case is that $1.3 trillion in semiconductor value was priced for perfection, and the labor market removed the possibility of a rate cut this year. Futures markets now almost fully price a quarter-point hike by December.
What changed in one week. On Monday, the market was operating on a simple thesis: the Fed holds, inflation cools, and AI earnings carry equities higher. By Friday, every element of that thesis had a crack. April CPI, reported last month, showed 3.8% year-over-year inflation, the highest since mid-2023. Brent crude traded between $91 and $99 in the past ten days. And a labor market that was supposed to be slowing just printed its strongest month in over a year.
Kevin Warsh chairs his first FOMC meeting June 16–17. It will include fresh economic projections and an updated dot plot. Before that, Tuesday’s May CPI print arrives. If it comes in above April’s 3.8%, the conversation shifts from “how long does the Fed hold” to “how soon does the Fed hike.” Prediction markets already assign 95% probability that year-over-year CPI will exceed 4%.
The question for anyone holding a 60/40 portfolio is arithmetic. A six-month T-bill yields above 4%. The S&P 500 sits 3% below its Monday record. If the consensus forward return on equities requires both earnings growth and multiple expansion, and the second half of that just got harder, the risk-free alternative is doing more work than it has in years.
The SOX didn’t fall 10% because one company missed on guidance. It fell because a 172,000-job print told the bond market that rate relief is off the table, and the most crowded trade on earth needed two days to figure out what that meant. Tuesday’s CPI will tell us whether the repricing is just starting.
|