The 10-Year Is at 5%. Hold 486 on SOXX.
US Stocks · Options · News · Views
The tape slipped again. Yields tightened the growth discount.
Tuesday's pressure was rates, not another chip crash. SPX closed 7,586, down 0.4%. The Dow fell 0.6%. The Nasdaq fell 0.8%. Only energy and materials rose among the 11 S&P sectors. Nine finished lower. The PHLX Semiconductor Index opened higher and closed weaker. SOXX closed 498.85 after a 496.58 low, still around 500. Monday already cut the multiple. Tuesday did not pop the cycle, and it did not reclaim the trendline. SMH closed 542.11. QQQ closed 704.54 after 703.64. NVDA closed 212.17, a small bounce that did not fill Monday's discount. META closed 670.24. AVGO closed 339.27, already inside 324 to 365.
The 10-year settled near 5.00%, tagged 5.04% in the session, and sits at a 2007-area high. Oil is still above $100, which keeps the inflation worry alive and tightens the discount rate on growth. The tape went defensive into Wednesday's decision. Futures put a 25 bp hike at 92.4%. A hold would shock and could bounce first. A hike as priced means the SEP — inflation, unemployment, and the rate path — is the read, not a rush to flatten the book tonight.
The slowdown table is not set. The stage is already down.
The weekend call to pace the frontier is not an industry consensus. NVDA's Jensen Huang told President Trump, on the record, that a slower development pace will not be allowed to happen. Trump treats AI as national security. The United States does not get to ease off, and data centers do not get to stop. META is not on the slowdown side either. Zuckerberg wants the build faster. China will not pause. Australia's prime minister put it bluntly: without a homegrown frontier model, the country is a petitioner, not a sovereign. ECB President Lagarde is warning Europe not to live on US models.
Labs can talk about a lighter throttle. Chips, platforms, and nation-states already pulled the stage down. The table is not set. Retail can turn faster. Follow the data. Until orders cancel and data-center work stops, hearing "pace" as cycle-end is early. The real downshift is rewritten contracts and capex, not a weekend essay.
The hike is almost fully priced. The path matters more than one print.
92.4% is close to fully priced. A surprise hold is a short-term positive. A hike as expected puts the focus on how Chair Warsh talks about what comes next. The new chair does not like giving much forward guidance. The dots are still here this meeting. They may go after a year-end rebuild of five committees. This packet still has three numbers that matter: the unemployment path, inflation, and the medium- and longer-run rate track. He can say less. The votes still put those figures on the table.
Deutsche Bank's history: once tightening starts, it rarely stops after one move. Through a tightening campaign the 10-year typically rises more than 100 bp. It is already 5%. Another 100 bp is 6%. That is not tomorrow's print. It is the DCF variable growth stocks hate if the path stretches. Unprofitable names and 80x, 100x, even 150x forward multiples get thinner when the discount rate rises. Do not take those tickets into a hiking path, and do not rerun the 2022 high-multiple growth tape.
Goldman looked at seven prior tightening cycles. SPX averaged −2% in the first three months after hikes began, then +9% over the next 12 months. 2022 was the exception. Rates set the near-term chop. EPS still sets the medium-term path. If the heavyweights hold earnings, the index does not have to break with the small names. An ugly tape is not an automatic kill of the longer thesis. That holds only while earnings estimates stay high and are not rolling over.
A 10% SPX drawdown is a correction. 7,000 is still inside the range.
Using 18x to 22x on the 2026 EPS base and 2027 growth, the market is already trading 2027. The 2027 SPX swing is about 6,986 to 8,539, midpoint 7,762. Spot 7,586 sits near that middle. A move to 7,000, even 6,900, is still the lower band if EPS is not cut hard. Below 6,900 starts to look cheap, and only if earnings are not rewritten. Wall Street calls a 10% peak-to-trough a correction. 20% is closer to a bear. A 10% drawdown shows up about once a year on average. 5% corrections are more common, three to five times a year.
Ten percent down from here lands near 6,800 to 7,000. 7,290 to 7,400 has not even printed, let alone the lower rail. Ten percent up is around 8,500. Up 10% and down 10% are symmetric in this frame. Both are still a swing. Mapping the floor in advance is so a real print is treated as an opportunity, not a panic. The level is not here. There is no need to flatten for a 10% that has not happened, and no need to pretend 7,586 is already a crash line.
486 is the SOXX line. Wait for Warsh to write the path.
QQQ beta runs about 1.2 to 1.5, closer to 1.4 this year. If SPX does 10%, QQQ is more likely to do 12% to 15%. 687 is the breakdown. Below that, 637 can still be a reasonable lower band, about 10%. Spot 704 is still a stretch from 687. The level is not here. Mark it. Do not turn a drawdown that has not printed into a trade.
SOXX lives 486 to 532, last 499. Through 532, then 577, 655 is in view, about 30%. A break of 486 changes the story. That is not a garden-variety dip. Watch 436 to 443 on the way. If the AI narrative is actually rewritten, the worst map is below 368, about 26%. That is a limit, not a target, and not a bid to catch now. SMH has not tracked SOXX this year. Weighting diverged. Last year the structures matched. This year they have been split for about six months. 534 is the key. Spot 542. If SOXX also breaks, SMH's worst map is below 421, about 22%. Upside 671 is about 23%. Same magnitude both ways. Wait for which side breaks.
AVGO is already inside 324 to 365, last 339. History is about 10.5x to 17x. This cycle is priced on 30x to 40x growth. If a name this strong will not hold 324, it reads as institutions losing confidence in the high-growth path. Mark the line. It is close. MSFT is stuck under 514. It cannot clear overhead, and support is nearby, so the trade window is thin. If the position is already cut, either buy a decline or wait for a breakout and follow the trend. Software is the same tape: a breakout creates a direction. Failure keeps it a range.
September's chop is not done. A tight two-way range with no direction is not a day-trade. Single names look dull because everyone is waiting on the same packet. Either SOXX breaks higher and the long is clear, or 486 gives and the complex is weak. Let Wednesday's SEP write the hiking path before deciding whether chips are still a range or already a trend.