The 10-year touched a 2002 high, then let go. The board took the room.

No pre-open or mid-day edition was published today, and no close has been published since Sep 23. This close is graded against the last thesis this desk put on the board.

The standing thesis · the split survived its own testThe Sep 23 close read that session as a split inside growth: chips sold and software bought while the 10-year hit 5.10%. It wrote down the test. If the 10-year kept rising and software rolled over to join the chips, Sep 23 was a lag, not a split. If software held with yields near or above 5.1%, the board was pricing something other than the rate. Today the 10-year touched 5.344% intraday, higher than on Sep 23, and software went 6 for 6. On that test, the split read held. The caveat is sampling: this desk is comparing two sessions a week apart and didn't publish on the days between, so "held" means it survived its own falsification test today, not that it's proven.

What led · software, and the chips came backSoftware went 6 for 6, averaging +2.62%: Unity +6.55%, Salesforce +3.10%, ServiceNow +2.80%, Palantir +1.60%, Datadog +0.95%, Snowflake +0.72%. UiPath added +3.74%. Semiconductors went 5 for 7, averaging +2.14%. That average is skewed by Wolfspeed at +10.81%, for which we pulled no company headline, so we aren't attributing a cause. Excluding Wolfspeed, the group averaged +0.70%: Micron +3.03%, Nvidia +1.09%, TSMC +0.66%, AMD +0.65%, against Intel −0.19% and Qualcomm −1.06%. Fintech and energy each went 3 for 3.

What lagged · health care and the heaviest weightsHealth care went 0 for 5, averaging −1.08%: Moderna −1.89%, Pfizer −1.40%, Gilead −1.04%, AbbVie −0.60%, Eli Lilly −0.49%. Mega-cap tech went 1 for 3 (Apple −0.81%, Microsoft −0.02%, Meta +0.10%). Cybersecurity, which led on Sep 23, cooled to 2 for 3, averaging +0.62%.

The read · the chips keep trading opposite the yieldIn each of the last three closes this desk published (Sep 17, Sep 23 and today), the chips have moved opposite to the 10-year. Software didn't need the yield to cooperate in either of the last two sessions. The defensive sleeve, health care, was the group that sold.

What would say this is wrongToday the yield retreated from its high, which makes a green board easier to explain. The cleaner test is a session where the 10-year closes at a new high. If software still holds on that day, the split is real. If it falls with everything else, today was a relief bid. That's an observation about today's structure, not a forecast about tomorrow's direction.

A 24-year yield high, a third carrier headed to the Middle East, a chip giant's record quarter and a bank succession move all landed on the same session. Every item below is attributed to a named outlet, and every market link is stated as observation rather than cause.

Rates · a 2002 high on the 10-year, then a turnCNBC reported that the 10-year yield hit an intraday high of 5.344%, its highest level going back to 2002, then turned around by late morning, with the 10- and 30-year yields last lower by 5 and 3 basis points (CNBC · Oct 1). The Motley Fool's midday headline read "Stocks Edge Lower as Treasury Yields Surge to 24-Year High," and TheStreet called the finish "flat on rising Treasury yields". Separately, The Motley Fool reported JPMorgan's Jamie Dimon saying hyperscaler AI spending could reach $1 trillion next year, a boom the piece says also worries him on inflation. The thread: stocks were lower at midday with the yield near its high, then finished higher as the yield came off. Software and the chips, the long-duration end of the market, were bid on a day the rate touched a 24-year high, but the bid showed up only after the yield had pulled back.

Energy & geopolitics · a third carrier and $100 BrentCNBC headlined that Brent oil jumped more than 4% as the U.S. reportedly sent a third aircraft carrier to the Middle East, and in a quarter-end piece wrote that stocks slowed in Q3 as the Iran war sent oil above $100. 24/7 Wall St. reported "Marathon Petroleum Climbs 5%, Valero Energy Gains 4% as Refiners Outrun Integrated Majors; Exxon Mobil Stays Flat," and Benzinga listed an ExxonMobil analyst who is "no longer bullish" among Thursday's top downgrades. The thread: energy went 3 for 3 on the board, averaging +1.21% (ConocoPhillips +1.53%, Chevron +1.42%, Exxon +0.69%). Exxon was the weakest of the three on the day it drew a downgrade. Our sentiment engine reads Exxon as neutral (4 positive, 0 negative across 15 headlines). Oil is also the input that bond markets have been pricing into inflation, which ties this item back to the first one.

Chips · a record quarter that didn't get the old reward, and a $150 billion buybackMicron reported after Wednesday's close, at 33.42 against a 31.77 consensus estimate, per our earnings calendar. Investopedia wrote "Micron Investors Aren't Rewarding Strong Earnings the Way They Used to." 24/7 Wall St. ran "Micron Stock Slips Despite Record Q4, Analyst Price Hikes" in the morning, and Benzinga reported "Micron's $150 Billion Contract Book Is Bigger Than Its Full-Year Revenue." The Motley Fool reported "Nvidia Authorizes a Record $150 Billion in Stock Buybacks." The thread: Micron, slipping in the morning headlines, finished +3.03%, the fifth-best name on the board, and Nvidia closed +1.09%. The reversal came on the same morning-to-afternoon turn as the yield. That's consistent with the chips trading opposite the 10-year, but one session is not a mechanism. Our basket reads both names as mixed to neutral across 15 headlines.

Banks, defense & people · a succession move, a stress-test rewrite, a first F-35The Motley Fool reported late Wednesday that JPMorgan elevated Doug Petno and Troy Rohrbaugh to co-presidents as the board accelerates succession planning. Zacks asked what the Fed's revamped bank stress tests mean for U.S. banks. A Lockheed Martin release via PR Newswire marked the arrival of Finland's first F-35. The thread: finance went 2 for 2 on the board (JPMorgan +0.71%, Visa +0.14%) after going 1 for 4 on Sep 23, so this time the banks moved with a high-rate tape that is conventionally read as helping them. Our sentiment engine reads JPMorgan and Lockheed as bullish on the label but neutral on 13 of 15 headlines each. That's a thin news signal, so we're not reading it as conviction.

What is actually being heldIf today's leaders are the book, it's software and chips that were bid while the 10-year retreated from a 24-year high. Today didn't test what they do if the yield closes at that high. If today's laggards are the book, it's health care and mega-cap weights that sat out a broad advance.

Three threads land before the next reactionThe September jobs report is due at 8:30 a.m. ET Friday, before the cash open, per CNBC. Morningstar previewed it as a hiring plateau after August's surge. A labor print lands directly on the yield that drove today. CNBC's carrier headline keeps the oil thread live overnight. Nike reported after the bell, and 24/7 Wall St. reported revenue of $11.21 billion against $11.32 billion expected, EPS of $0.48 against $0.44, and shares down about 5% after hours. We're describing when news arrives relative to when a position can act on it, not calling a direction.

Weekend nextToday is Thursday, so tonight's gap is one session. But Friday's close hands straight into a weekend with an active Middle East thread. Whatever the jobs report does to yields gets held for two extra days before anyone can react to the geopolitical carry.

Estimates shown are consensus figures from our earnings calendar, not this desk's projections. The earnings calendar is nearly empty into Friday, so the jobs report is the marquee print. Earlier today, Accenture (3.29 vs 3.18 est), McCormick (0.86 vs 0.755) and Acuity (5.77 vs 5.57) reported above consensus before the open, and Hub Group (−0.04 vs 0.205) came in below it, per our calendar feed.

Overnight risk vetoThe input that drove today gets a new data point before the open. The jobs report lands at 8:30 a.m. ET, an hour before US equities can respond. The software-and-chips leadership has not yet been tested on a session where the 10-year closes at a new high. Define the level that says your thesis is wrong before the open, not after.