GOOD MORNING, Thursday split the market in two. June payrolls rose by just 57,000, well below expectations, cutting near-term Fed-hike odds and pushing the dollar to its biggest two-month drop. The Dow jumped to a record close. The Nasdaq still fell 0.8% because semiconductor stocks dropped another 5.4%. Meta’s plan to sell excess AI compute had turned a theoretical concern into a market question: are hyperscalers building faster than they can monetize? The macro backdrop just got friendlier. The AI hardware trade still has to prove that capacity and returns can grow together. MARKETS | TLDR
LABORThe Fed Just Got a Reason Not to RushThe U.S. economy added just 57,000 jobs in June, far below the 110,000 increase economists surveyed by Reuters had expected. Payroll gains for the prior two months were also revised lower, while the unemployment rate held at 4.2%. The report gave investors the clearest evidence in weeks that the labor market was beginning to cool. Markets immediately reduced the chance of another rate increase. The implied probability of a September hike fell to roughly 55% from 64.1%, and the dollar dropped 0.52%, its steepest two-month decline. That helped economically sensitive and rate-exposed parts of the market, with the Dow jumping more than 1% to a record close. But the reaction also exposed a split inside equities. Softer jobs normally support long-duration growth stocks because they lower the expected discount rate. This time, semiconductors kept falling. That means the chip selloff was no longer just a macro story. Lower rate pressure helped the market broadly, but it could not resolve concerns around AI positioning, valuation and the returns on the capacity being built. AI CAPACITYMeta Turned “Too Much Compute” Into a Market QuestionMeta is developing a cloud business to sell excess AI computing capacity, according to a Bloomberg report cited by Reuters. The strategy is still evolving, but the idea is to let developers access Meta’s models and pay for the compute needed to run them — effectively turning spare infrastructure into a commercial product. The market reaction made the read-through clear. Meta surged after the report, because selling unused capacity could improve the economics of its enormous AI buildout. CoreWeave and Nebius, two neocloud providers that count Meta as a major customer, fell sharply because the move could both reduce Meta’s spending with them and create a new competitor. For the chip trade, the bigger question is whether hyperscalers are building faster than their own workloads can absorb. Meta could simply be monetizing prudent excess capacity. But after Big Tech committed more than $700 billion to AI infrastructure this year, the market is starting to distinguish between “more compute” and “more productive compute.” That distinction helps explain why semiconductors fell even as the macro backdrop improved. HEADLINES
UPCOMING
DEEP INSIGTHSDollar Slides After Jobs Data, Chipmakers Weigh on StocksRead this for the day’s central divergence. Softer payrolls reduced Fed-hike expectations, weakened the dollar and supported the broader market, yet semiconductor stocks still fell hard. That disconnect shows why the AI selloff had become increasingly independent of the macro rate story. Meta Is Building a Cloud Business to Sell Excess AI CapacityThis is the structural question underneath the chip selloff. Meta’s plan could improve returns on infrastructure that would otherwise sit idle, but it also suggests the AI buildout has reached a scale where spare capacity itself needs to be monetized. For investors, that shifts the focus from how much compute gets built to how efficiently that compute earns a return. |