Opinion · Portfolio Construction · September 10, 2026
A Single Rate Number Moved Eight of My Holdings the Same Direction Today
Micron, SanDisk, AMD, Intel, Vertiv, CoreWeave, Nebius and its own leveraged ETF all fell Thursday, detailed in full in today’s News piece. Not one of them had a company-specific headline behind its decline. A wholesale-inflation print of 5.4% and crude oil crossing $106 a barrel pushed Fed rate-hike odds to 70% for September, and that single macro fact moved a memory chipmaker, a leveraged ETF, an AI cloud provider and a data-center cooling company the same direction, by similar magnitudes, on the same day. I hold all eight. That is not eight independent theses being tested at once. It is one factor bet wearing eight different tickers.
Figure 1 · Nine names, September 9 to September 10 close
Eight names moved the same direction on the same day; only Apple, with its own real catalyst, moved the other way
Nine names from this site's coverage list, September 9 to September 10, 2026 close, each indexed to its own starting price. Apple is the one name with a same-day, company-specific catalyst; the other eight had none. Source: Massive Market Data.
1. A sector label describes a business. It doesn't describe who's selling it.
By nominal sector, Intel and AMD are semiconductor makers, Micron and SanDisk are memory makers, CoreWeave and Nebius are AI cloud infrastructure, Vertiv is data-center cooling and power equipment, and NBIL is a leveraged derivative on Nebius alone. Four labels, arguably five businesses with genuinely different customers, cost structures and competitive positions. None of that mattered Thursday. What mattered was that every one of them gets bid up on the same "AI infrastructure demand is durable, rates stay low enough to fund it" story, and sold on the same "rates might be higher for longer" story — which is exactly what a hot PPI print and a 70% Fed-hike probability delivered. The label on the ticker is a description of the business. The correlation is a description of who owns it and why they're all reacting to the same input.
2. This is the same question my July capex work was asking, from a different angle
Where this connects to my own thesis
Back in July, I asked whether AI infrastructure spending was still paying for itself, after a session where TSMC's capex raise, an IBM warning and a Netflix guidance miss all landed the same day. I later went and checked with real numbers in The AI Capex Reality Check: Alphabet generated $0.28 of free cash flow for every $1 of Q1 2026 capex, and Meta generated $0.65 — neither company self-funding its current build-out pace from cash alone. Today's selloff isn't that question again. It's the adjacent one: even if the capex-to-cash-flow gap never widens, this entire cohort is still exposed to a second, independent risk — that the multiple the market is willing to pay for that spending compresses whenever financing conditions get more expensive. Thursday, the trigger wasn't a capex number. It was a rate number. The stocks didn't seem to care about the distinction.
3. What it would take for the other 24 positions to offset this basket
The eight correlated names above are worth $11,666.49 combined at today's close — 35.0% of my $33,318.75 book. The table below asks a different question than today's Letter: not what the book is worth under further moves in this basket, but how much the other 24 positions, worth $21,652.26 combined, would need to move in the opposite direction just to net the book flat against a repeat of today's shock.
| If the 8-name basket moves… | Dollar swing | Offset needed from the other 24 positions |
|---|---|---|
| Unchanged | $0.00 | — |
| Another SMH-sized move (−2.44%, today's semiconductor-index decline) | −$284.66 | +1.31% |
| Another SLV-sized move (−5.30%, today's silver decline) | −$618.32 | +2.86% |
| Illustrative stress case (−10%) | −$1,166.65 | +5.39% |
| Illustrative rebound (+5%) | +$583.32 | −2.69% |
Assumptions, stated plainly. The $11,666.49 and $21,652.26 figures are my own real position values at today's close, from my own share counts. Two of the five move-sizes reuse today's own real, dated SMH and SLV index changes as illustrative magnitudes only — not a claim that either instrument leads or predicts this specific basket. The 10% stress case and 5% rebound are not sourced to any analyst estimate; they exist to show the offset arithmetic at round numbers, in both directions, rather than assuming losses only run one way. This is not a hedging recommendation and not a forecast of what the other 24 positions will actually do; it's a measure of how much independent movement I'd need from the rest of the book to call this basket's risk diversified away, and right now that number is not small.
What this changes, and what it doesn't
WatchTwo things that would tell me this framework needs revisiting
1. Whether Friday's CPI print comes in hot, in line, or cool relative to expectations. Both Fool pieces cited in today's News article flagged CPI as the next data point the Fed weighs before its September meeting; a genuinely surprising print in either direction is the next real test of whether this basket keeps trading as one factor. 2. Whether any of these eight names decouples from the group on its own news before then. If Micron or CoreWeave reports something company-specific that moves it independently of the other seven, that's evidence the correlation is looser than one session suggests; if the whole group keeps moving together through earnings season, the concentration this table quantifies is a standing feature of the book, not a one-day artifact.
I hold all eight names in this basket and made no trades Thursday. The read I'm taking isn't that a rate-driven, multi-name selloff is unusual — it isn't, and I've logged versions of it before. It's that 35% of my book just demonstrated, in one session, that it responds to a single macro input rather than eight independent ones, and that's a fact about my portfolio construction I hadn't quantified in dollar terms until today.