The 10-Year Hit Its Highest Since 2007. What Was Left Standing?

The indices: three directions the day before, one direction now

All three indices closed lower on Wednesday: the $NASDAQ(.IXIC)$ Composite fell 1.13 per cent to 26,936.04, down 308.24 points on the day, giving back the record it had set in the previous session; the $S&P 500(.SPX)$ fell 0.75 per cent to 7,706.03, down 58.61 points; and the $Dow Jones(.DJI)$ fell 0.68 per cent to 51,511.59, down 352.10 points.

The previous session had the three running separately — the Nasdaq closing at a record, the Dow lower, the S&P 500 finishing flat to within 0.06 points. A day later they were back on the same heading, and all of it downward. What can be established here is the direction, not the make-up: without the day's advancers and decliners and the contribution by sector, three closing levels cannot tell you whether this was broad selling or a few heavyweights doing the pulling. Two paths follow. Should the three keep moving together, the whole market is repricing off one variable and reading each index on its own no longer works. Should they separate again soon, Wednesday was one shared trim rather than a new theme. The thing to watch is whether the Dow's list of fallers spreads out — once it is no longer concentrated in financials and networking, what is being cut is overall exposure.

Rates: what pulled the three lines back together was one number

The data came first. S&P Global's flash composite purchasing managers' index for September came in at 58.4, against 56.0 in August, the strongest reading since July 2021; the flash services index came in at 58.7, where the market had expected 56.0. After that, the 10-year Treasury yield closed at 5.11 per cent, 15 basis points above the previous day's 4.97 per cent and the highest since July 2007; the two-year rose 13 basis points to 4.9 per cent, and the five-year touched 5 per cent during the session, the first time since 2007.

CME FedWatch had the market's odds of an October rate rise at 73 per cent that day, against 53 per cent the day before. The Federal Reserve had raised rates on 16 September, taking the target range to a band of 3.75 per cent to 4.00 per cent. So the chain ran backwards on Wednesday: the better the data, the higher the odds of another rise, the higher the risk-free rate, and the higher the bar equities have to clear. The same split applies. Should September's reading prove to be a one-off, with October's jobs and price data failing to follow, yields come back down first and Wednesday counts as an overreaction. Should the data that follows stay strong, a risk-free rate above 5 per cent is here to stay, and anything priced off distant cash flows has to be worked out again. What to watch is October's releases, and whether the five-year can hold above 5 per cent.

Memory and chips: the ones that gained most the day before fell most now

Memory again fell the hardest: $SanDisk Corp.(SNDK)$ closed 3.73 per cent lower at US$1,816.57, $SK hynix(SKHY)$ 3.12 per cent lower at US$189.28 and $Micron Technology(MU)$ 2.22 per cent lower at US$1,071.88. Those three had been the three largest gainers in the previous session. The chip end did not hold either: Nvidia closed 1.47 per cent lower at US$225.51, AMD 1.47 per cent lower at US$614.61 and Intel 1.02 per cent lower at US$122.60.

Whether the chip names could keep up got its answer on Wednesday: they did not, the whole chain stepped back together, and the end that had gained most was the end that fell most. That says the previous session was not money moving out of chips and into memory; both ends are entries in the same book, and when rates move they move together. What has to be kept apart is that the price fell, not the agreements already signed — a contract settles how much SanDisk collects in financial year 2028, not the rate at which that money is discounted back to today. The thing to watch is still Micron's results after the close on 30 September: should it lift both its gross margin and its guidance for the coming quarter, this fall can be separated from the fundamentals; should the guidance come in merely flat, the market has to work out the rate side and the fundamental side at once.

Meta: everything was falling and it rose

Few large companies closed higher on Wednesday, and $Meta Platforms, Inc.(META)$ was one of them, up 1.02 per cent at US$744.10. Meta Connect opened that day, with a US$1,299 pair of VR glasses unveiled and a way for the Muse AI assistant to make money: a small fee on every transaction it completes. The list of partners announced alongside it includes Walmart, Best Buy, Sephora, Expedia and Instacart.

A list like that was exactly what Muse had been missing, and now there is not only a list but a place where the money comes in, with the size of the cut still unstated. $Alphabet(GOOG)$ closed 3.58 per cent lower at US$334.98, a fall more than three times the Nasdaq's; reporting has tied that drop to early adoption of Muse, citing Apptopia data that puts downloads above 2.8 million in the first 12 days. That can only account for part of it — almost every large technology name fell that day, and the rate side and the competition side cannot be told apart. This line splits two ways as well. Should the list keep growing once the size of the cut is published, Muse is genuinely starting to earn its own money rather than just gathering usage. Should the list stop at these names, the fee is only a sentence. What to watch is the size of the cut, and which companies announce next that they are plugging in.

SpaceX: last week's contract could not hold off this week's unlock

$SpaceX(SPCX)$ closed 4.11 per cent lower at US$148.36. It had just won a contract with a visible horizon: on 18 September NASA bought three more crewed Dragon missions, worth up to US$946 million, with a period of performance running through 2030, which takes the cumulative crew transport award to somewhere around US$5.92 billion.

Public reporting has a tranche of restricted shares unlocking on 24 September, alongside reported insider selling plans, and that supply sits on the other side of the contract. Other reporting says the company cannot qualify for the S&P 500 before June 2027, and that continuing losses could push that date further out, so there is no help coming from index money for now. The memory section works the same way: a contract signed through 2030 can pin down future revenue, but not how much stock is coming up for sale today. Two paths again. Should the price recover once this tranche has been absorbed, the pressure really was only supply. Should it keep sliding after that, the problem is the valuation itself and has nothing to do with the unlock. The thing to watch is trading volume in the sessions right after the unlock.

Side by side: terms govern revenue, not the rate

On Tuesday the market was willing to pay for things with a date, with terms, that you can look up. On Wednesday not one of those things had changed — SanDisk's capacity is still locked in, NASA's contract is still signed, Muse's list of partners is still growing — and they fell together anyway. So the judgement about terms does not have to be withdrawn, only added to: terms settle how much a company will collect, not the rate at which that money is discounted back to today. What changed on Tuesday was what the money was willing to pay for; what changed on Wednesday was what today itself is worth.

The above is personal analysis, not investment advice.

💬 【Talking Point】

$Meta Platforms, Inc.(META)$ has not said what cut Muse takes on a transaction. Until that number is published, does a fee on completed orders change how you value the assistant at all?

💰 【Bounty】

Drop your view in the comments and there are Tiger Coins in it for you! 🎁

🔔 Better shared than saved — tag a friend and split the coins!