Jobs Missed. The Dow Hit a Record. Chips Didn't Care.
GOOD MORNING, U.S. markets were closed Friday for the Independence Day holiday, so the last tradable signal came from Thursday — and it was unusually split. June payrolls rose by just 57,000, roughly half what economists expected, cutting the odds of another near-term Fed hike and sending the Dow to a record close above 52,900. The Nasdaq still fell 0.8% because semiconductor stocks dropped another 5.4%. Meanwhile, oil finished Friday near pre-war levels as U.S.-Iran talks and rising Gulf production eased supply fears. Macro conditions improved. The chip trade still had its own problem. MARKETS | TLDR
LABORThe Jobs Miss Gave the Fed Room to WaitU.S. employers added just 57,000 jobs in June, well below the 110,000 economists surveyed by Reuters had expected. The unemployment rate held at 4.2%, while May payroll growth was revised lower. Markets immediately reduced expectations for additional Fed tightening, with the implied probability of a September hike falling to about 55% from 64.1%. That mattered because the market had spent June worrying that resilient growth and sticky inflation would force the Fed to tighten again. A softer employment report gives policymakers more room to wait and observe whether the earlier energy shock continues to fade. The Dow responded with a record close, helped by a broader rotation into economically sensitive and non-tech names. The important caveat is that this was not a full risk-on reaction. The Nasdaq still fell because semiconductor stocks remained under pressure. Easier rate expectations helped the broader market, but they were not enough to repair a sector wrestling with positioning, valuation and questions about how efficiently the AI buildout is being monetized. AI CAPACITYThe AI Trade Is Starting to Separate Capacity From ReturnsMeta is developing a cloud business designed to sell excess AI computing capacity, according to a Bloomberg report cited by Reuters. The idea is to monetize infrastructure that might otherwise sit underused while giving developers access to Meta’s AI models and compute, similar to offerings from the major cloud platforms. The read-through matters because hyperscalers are spending enormous amounts on chips, servers and data centers. If Meta can rent unused capacity, that improves the economics of its infrastructure. But it also raises a more uncomfortable question for the broader AI hardware trade: are some buyers building more capacity than their own internal workloads can immediately absorb? That helps explain why semiconductor stocks can sell off even while AI demand remains structurally strong. The market is becoming more selective about the difference between adding compute and earning an attractive return on that compute. Meta’s plan may be financially rational for Meta while creating new pressure for neoclouds such as CoreWeave and Nebius — and a new way for investors to think about excess capacity across the AI stack. HEADLINES
UPCOMING
DEEP INSIGTHSTrading Day: Chips Are Down, and So Are PayrollsRead this for the cleanest cross-asset framing of the holiday-shortened week. Softer payrolls pulled the dollar and rate expectations lower, but chip stocks kept falling anyway. The divergence shows why the AI selloff had become more than a simple interest-rate story. Meta Is Building a Cloud Business to Sell Excess AI CapacityThis is the deeper structural question underneath the semiconductor weakness. Meta’s plan suggests hyperscalers may increasingly treat compute capacity as an asset to monetize externally rather than something used only for internal workloads. That can improve returns for the platforms while changing the competitive landscape for neoclouds and the assumptions investors make about future hardware demand. |