Interest Costs Jump, SOXX Still Capped
US Stocks · Options · News · Views
Core ran hot. Hike odds hit 86.5%
The four major indexes closed green on September 11 in New York. Most sectors finished higher. Communication services, tech, and discretionary led. Utilities, healthcare, and energy lagged. There was no new tape story. The session priced three prints: core inflation, Michigan sentiment, and the August Treasury statement.
August core CPI rose 0.3% month over month versus 0.2% expected. Year over year it was 2.4%, in line, down from 2.5%. Headline rose 0.4% month over month, in line, much faster than the prior 0.1%. The year-over-year rate stayed 3.4%. Super-core jumped from 0.2% to 0.5%. Core services ticked up. Core is still cooling on a 12-month basis. The hot month is in services after you strip shelter.
CME odds of a 25 basis point hike next Wednesday rose to 86.5%. The tape is treating a hike as the base case. Interest expense is already this heavy. Another step makes the fiscal ledger worse. Both ends are tight. This is not “inflation cleared, so policy eases.”
Michigan cracked. Inflation expectations lifted again
The preliminary Michigan sentiment index printed 47.8 in September, down from 51.7, versus about 51 expected. Only May’s 44.8 has been lower in the survey. The damage is in expectations: that index fell from 51.5 to 45.8, under the about 50.5 estimate. Current conditions only slipped from 51.9 to 50.9. Households are not saying today got suddenly worse. They are saying the path ahead did.
One-year inflation expectations rose from 4% to 4.6%, the highest since June. Five-year expectations rose from 3.3% to 3.4% after three months stuck at 3.3%. When oil is elevated, inflation expectations move first. Hikes can lean on demand. They do not pin crude. Expectations then bounce between those two constraints. Even a hold next week does not take later hikes fully off the table.
August’s deficit looks small. Interest is already expensive
Treasury released the August budget in the afternoon. Receipts were $360 billion, a bit above $344 billion a year earlier. Outlays were $527 billion versus $689.1 billion a year ago. Calendar gaps and tariff payment timing pulled the outlay print down. August 1 was a Saturday, so the usual early-month Medicare and Social Security checks hit in July. July’s deficit was $432 billion. August’s was $166.8 billion. That gap is mostly calendar, not a sudden fiscal heal. Strip the timing noise and August is about $248 billion, a touch worse than last year.
Over the past six months, outlays have a steeper slope than receipts. Receipts are still stuck in a narrow band. Outlays are climbing toward the pandemic highs. The gap between the two lines is widening. The fiscal 2026 deficit through 11 months is $1.97 trillion, already even with last year, with one month left. On this pace, the full-year gap may run about $200 billion wider than last year and rank as the third-largest on record, behind only 2020 and 2021. Those were crisis years.
Interest is tighter. August interest outlays were $98 billion. With one month left in the fiscal year, the first 11 months are already $1.267 trillion, up about 13% year over year. On a trailing 12-month basis the bill is about $1.4 trillion, a record. Social Security is about $1.66 trillion on the same rolling window and growing more slowly. On that slope, interest can pass Social Security before the end of 2028. Debt can be rolled. The longer it rolls, the more interest crowds out other spending, and the more new issuance fills the hole. A hike next Wednesday hits Treasury yields first. Small books can turn. Long-term capital can live with a drawdown. It cannot live with one or two years that hand back both principal and gains and rewrite a retirement plan. Follow the tape. Do not underwrite Treasury’s choice.
SOXX closed 527. The trendline still caps it
SOXX rose 1.86% and closed about 527, high about 531. The falling trendline is still overhead. The line drifts lower, so 527 is still a stall, not a break. After OpenAI’s new model last week, the group surged last Friday, chopped for four sessions, and tagged the door again today. The breakout window is still there. The closing confirmation is not.
Fundamentals and technicals can split in the short run. Rich can get richer. Cheap can get cheaper. Over a longer stretch it still mean-reverts. After this drawdown, some components have come back from rich to more reasonable. AMD is less expensive after the earnings reset. AVGO and MRVL sit in the same bucket. The group is not especially expensive here. If the close gets through the line, the tape still has a trade. Two later nodes: this year’s fourth-quarter prints, and whether OpenAI and Anthropic list, and when. Until a prospectus is out, it is talk. Listing into a washout draws more blood. Listing after the tape stabilizes can hold the heat.
A higher close next week turns the technical picture. Another weak open keeps compressing the range until a direction is chosen. Until it clears, wait. The fundamentals can be watched. The short-term switch is still the close.
Capital Economics’ five warnings. Not this year. Medium-term poor
Check the AI book each quarter. The nearer window still looks relatively safe for one to two quarters. The latest node is this year’s fourth quarter, when 2027 capex gets disclosed. Some names may speak as early as the third quarter. Until that node, do not write a bubble burst as a fact.
Capital Economics is a UK independent shop. They list eight equity-bubble gauges. Five already sit at extremes: valuation, earnings expectations, index concentration, equity issuance, and foreign ownership of US stocks. That matches prior peaks. The S&P 500 forward PE is about 21 times, down from about 23 at the end of 2025, and cheaper versus other markets. 1999 to 2000 also looked less expensive first. Cut the forward earnings path and the multiple suddenly looks very high. Consensus EPS for the next 12 months is already near the dot-com peak. Most of that growth is long-duration and clustered in tech, which makes it more brittle. The largest weights are all tech-tied. Concentration needs no new speech. Issuance is the IPO tape. If OpenAI and Anthropic actually list, that gauge goes more extreme. The 2021 small-cap and shell-merger melt-up then wipeout is the same late-cycle shape. Foreign buying of US equities also looks like prior peaks.
Three gauges are not extreme. Company fundamentals have not deteriorated badly. Volatility is pinned by zero-day options, so VIX does not swing like it did before that product took over. Leverage is not manic. This is not Korea stacking margin until it breaks. Hedge funds are still in low-exposure watch mode, not stacking leverage into price. Five are there. Three are not. Their call is that this year probably does not break hard, and the medium-term outlook is poor, roughly next year through mid-to-late next year. That is a reference, not an order to flatten the book today. The view updates with data and nodes. If something severe shows up, it belongs in the title.
FOMC is next week. The index has not chosen a large trend. Watch whether SOXX closes through the line. Fiscal and interest costs are already on the table. A hike is one more step of cost. Near-term switches: SOXX falling trendline, only a close counts; hike odds already at 86.5%; trailing interest about $1.4 trillion, so another hike asks the bond market first. If it does not clear, wait.