GOOD MORNING, Wall Street heads into the new week with two macro pressures suddenly looking less severe — and one AI problem refusing to go away. June payrolls rose by just 57,000, cutting the odds of a July Fed hike to below 20%. Oil also finished the holiday-shortened week near pre-war levels as U.S.-Iran talks made incremental progress on Hormuz shipping. Yet semiconductor stocks kept selling off, with the Philadelphia Semiconductor Index down another 5.5% Thursday after a 6.3% drop the day before. Easier rates and cheaper oil helped the broader market. They did not fix the AI trade’s expectations problem. MARKETS | TLDR
LABORThe Fed Just Got More Room to WaitU.S. employers added only 57,000 jobs in June, about half what economists had expected, while May’s payroll gain was revised down to 129,000 from 172,000. The unemployment rate edged lower to 4.2%, but that partly reflected a decline in labor-force participation rather than an acceleration in hiring. Markets treated the report as a clear reduction in near-term tightening risk. Fed-funds futures pushed the probability of a July rate hike below 20%, down from much higher levels earlier in the week, while the implied chance of a September hike fell to around 60% from roughly 75% before the report. That matters for equities because the market had spent much of June worrying that sticky inflation and strong growth would force the Fed to tighten again. A softer labor market removes some urgency. But it does not automatically create a bullish growth-stock setup: semiconductor shares still sold off hard despite the rate relief, showing that AI valuation and positioning concerns had become their own driver. AI CAPACITYThe Market Is Starting to Ask Whether AI Has Too Much CapacityMeta’s reported plan to build a cloud business around excess AI computing capacity triggered an unusually revealing market reaction. Meta itself surged as investors welcomed a potential way to monetize infrastructure that might otherwise sit underused. But neocloud providers CoreWeave and Nebius fell sharply, and the broader semiconductor complex continued to weaken. The read-through is uncomfortable for the AI hardware trade. Meta is spending aggressively on compute, but the idea that it may have enough excess capacity to sell externally raises a question investors had not needed to ask during the first phase of the boom: are hyperscalers building faster than internal demand can absorb? That does not mean AI demand is collapsing. The spending plans remain enormous, and selling spare capacity could simply improve returns on infrastructure that was always intended to be flexible. But after a record-breaking second quarter for semiconductor stocks, the market is beginning to distinguish between “more AI capacity” and “more economically productive AI capacity.” That distinction could define the next leg of the trade. HEADLINES
UPCOMING
DEEP INSIGTHSFed Seen Less Likely to Raise Rates as Job Growth SlowsRead this for the policy shift behind Thursday’s cross-asset move. The June jobs report did not signal a recession, but it materially reduced the urgency for another Fed hike. That distinction matters because the market can receive rate relief without receiving a growth scare — a theoretically positive setup for equities that semiconductor stocks still failed to capitalize on. Meta Is Building a Cloud Business to Sell Excess AI CapacityThis is the more interesting AI story beneath the chip selloff. Meta’s plan suggests hyperscalers may increasingly monetize infrastructure externally when internal workloads do not fully absorb it. That can improve returns for Meta while creating new competition for neoclouds — and it raises a broader question about whether the AI buildout is approaching a point where capacity growth starts to outrun captive demand. |