The Family Fight

Warsh got the three dissents he asked for, the bond market read a Fed falling behind, and the Dow fell 1,153 points — its worst day since April 2025. Then Microsoft and Meta reported, and split the AI trade cleanly in two.

Kevin Warsh said he wanted a good family fight. He got one. The Federal Reserve held rates for a fifth straight meeting, but three regional presidents dissented in favor of a hike — the most one-directional dissent since 2016 — and the bond market read the whole thing as a central bank falling behind. The Dow fell more than 1,100 points, its worst day since April 2025. And then, after the bell, the two biggest reports of the night split the AI trade cleanly in two.

The hawkish hold

The FOMC left the funds rate at 3.50–3.75% on a 9–3 vote — the fifth consecutive hold. What made it notable was the shape of the opposition. Cleveland's Beth Hammack, Minneapolis's Neel Kashkari and Dallas's Lorie Logan all dissented in favor of an immediate quarter-point hike, the largest bloc of one-directional dissents since September 2016. The statement stayed short, in keeping with Warsh's stated disdain for forward guidance, and described activity as expanding at a solid pace despite the uncertainty from the Middle East conflict.

Warsh, who has called inflation a choice, leaned into the discord rather than smoothing it over. Asked about the split at his press conference, he framed it as exactly what he wanted from the committee.

The market did not find it reassuring. Stocks briefly turned higher during the press conference, then slid into the close. The next real test is September 15–16; before then, Warsh speaks at Jackson Hole on August 27–29, which is now the single most-watched date on the calendar for any hint of where this committee is actually headed.

The bond market's verdict

The decision was a hold, but the reaction was a repricing at the long end. The 30-year Treasury yield jumped roughly ten basis points to above 5.2%, a level not seen since 2007, and the 10-year rose about seven basis points to above 4.67%. That is the tell: when the Fed holds and the long bond sells off, the market is not pricing easier policy — it is pricing a central bank that may be letting inflation run. With crude up more than 20% for July, that fear has a live catalyst.

Rising yields, not the decision itself, did the damage in equities. The Dow's 1,153-point drop was led by a 6.8% fall in Caterpillar, and the Nasdaq closed about 9.8% below its early-June record, on the brink of a correction. The chip complex, already in its own private bear market, kept sliding — and in Asia, the Kospi triggered a circuit breaker for a second straight day.

Then the receipts came in — and disagreed

After the close, Microsoft and Meta reported within minutes of each other and told opposite stories about the same AI boom. Microsoft delivered fiscal fourth-quarter revenue of $90.01 billion against a $87.62 billion estimate, with Azure growing 43% in constant currency — well above the roughly 40% Street bogey — and Azure revenue crossing $100 billion for the fiscal year for the first time. Crucially, capital expenditure came in near $35.8 billion, below the $36 billion-plus the market feared, and management guided Azure to accelerate to 45%. The stock jumped about 9% after hours.

Meta ran the other way. Revenue rose 28% to $60.80 billion, but earnings per share of $6.18 missed by more than a dollar, net income fell 14%, and the operating margin collapsed from 43% to 31%. Then management raised the floor of its 2026 capital-expenditure range to $130 billion. Same arms race, very different bill. Shares fell about 9%.

This is the refinement of last week's lesson. When Alphabet and Tesla sold off on capital spending, the takeaway looked like a blanket verdict against AI investment. It wasn't. Microsoft spent heavily and was rewarded because the spend arrived with acceleration; Meta spent heavily and was punished because it arrived with margin erosion. The market has stopped asking whether these companies are spending and started asking what the spending buys. That is a more discriminating tape — and a healthier one.