Opinion · September 2026
Same Rate Hike, Ten Different Reactions: What Wednesday's Split Tape Actually Priced
The Federal Reserve raised its target rate on Wednesday for the first time in three years, and the broad market did what a broad market is supposed to do on a hawkish surprise: the S&P 500 fell, the Dow fell more, and the dollar and volatility both rose. That much is a clean, single-cause story, covered in full in today's News piece. Underneath it, ten names in AEA's coverage universe moved by more than 3% in either direction on the same day, and when I sorted them by why they moved rather than by how much, they split into three groups that behave nothing alike.
Figure 1 · Ten names, sorted by move and by whether a dated catalyst was found
The biggest moves of the day split cleanly into three different kinds of explanation
Circle position is the real percent move; row is a judgment call about whether a dated, sourced news item explains it, made using the reporting cited in today's News article. Green marks an up move, red a down move — color encodes direction, not the row classification. Source: Massive Market Data.
Three groups, three different things to believe
The first group is the one the headline is about: the broad index and crude oil, which fell because the Fed did what it did. The S&P 500's 0.44% decline and the Dow's 1.15% decline do not require a company-specific explanation — they are the market's aggregate response to one macro event, and they should be read as exactly that and nothing more.
The second group has a real story, and it is a good one to have: Intel rose 4.03% and Marvell rose 3.61% on a reported SK Hynix manufacturing arrangement, and Coinbase fell 4.42% against a backdrop of reporting that a competitor's token may soon win U.S. regulatory approval for a product Coinbase does not yet offer at scale. All three moves are large, and all three have a dated, sourced, company-specific reason behind them that a reader can go verify.
Priced Wednesday on a dated catalyst
Intel (+4.03%), Marvell (+3.61%) and Coinbase (−4.42%) all moved on reporting specific enough to name, date and link — two on the same SK Hynix report, one on regulatory context from the evening before.
Priced Wednesday on no catalyst I could find
SpaceX (+5.15%), Applied Digital (+4.14%), Super Micro (+3.40%) and CoreWeave (+3.00%) all moved by more than 3% with no dated, company-specific news item behind them in the sources checked for this piece.
That third group is the one I keep coming back to. Four separate AI-infrastructure-adjacent names, each moving 3% or more on a day when the dominant macro headline was a rate hike that should, if anything, make speculative growth names more sensitive to discount-rate pressure, not less. The most honest description of that group is that it moved on sentiment — a general willingness to bid up anything AI-infrastructure-shaped — rather than on anything specific to any one of the four businesses. I am not claiming that sentiment is wrong. I am claiming it is not the same kind of evidence as a signed SK Hynix agreement, and treating the two as equivalent is how a reasonable thesis turns into a story someone tells themselves.
| Ticker | Day | Group | Basis |
|---|---|---|---|
| SpaceX (SPCX) | +5.15% | Momentum | No dated catalyst found |
| Applied Digital (APLD) | +4.14% | Momentum | No dated catalyst found |
| Coinbase (COIN) | −4.42% | Catalyst | Hyperliquid regulatory read-through |
| Intel (INTC) | +4.03% | Catalyst | Reported SK Hynix Ohio talks |
| Marvell (MRVL) | +3.61% | Catalyst | SK Hynix partnership read-through |
| Super Micro Computer (SMCI) | +3.40% | Momentum | No dated catalyst found |
| Crude Oil (USO) | −3.52% | Macro | Broad commodity reaction to the hike |
| CoreWeave (CRWV) | +3.00% | Momentum | No dated catalyst found |
| Dow (DIA) | −1.15% | Macro | Broad index reaction to the hike |
| S&P 500 (SPY) | −0.44% | Macro | Broad index reaction to the hike |
Where this connects to my own thesis
This is the same live question I raised in yesterday's piece about the AI trade splitting into compute suppliers and compute users, and the same underlying question my Physical Limits of Compute whitepaper tried to get ahead of: is this cycle's spending still tracking demand that shows up in filings and signed contracts, or is some of it now trading on the expectation of demand that hasn't been confirmed yet? Wednesday isn't proof of the second thing. But a day where four infrastructure-adjacent names rally 3–5% with no dated catalyst, on the same session a genuinely sourced catalyst produced a comparable move at Intel, is exactly the kind of day that question should be asked about.
What it would mean if momentum-only names start behaving like catalyst names
I don't have a sourced forecast for where SpaceX, Applied Digital, Super Micro or CoreWeave go next — nobody published one today that I found. What I can do honestly is show what continued moves of Wednesday's shape would do to a name in that momentum group if they simply repeated, since ServiceNow, a growth-multiple holding of mine that fell 1.47% on the same session, gives a clean real number to build the arithmetic from.
| ServiceNow (NOW), from Wednesday's $139.82 close… | Implied price | My 6-share position | vs. today |
|---|---|---|---|
| Unchanged | $139.82 | $838.92 | — |
| One more session like today (−1.47%) | $137.76 | $826.56 | −$12.36 |
| Three sessions like today, compounded (illustrative) | $133.70 | $802.20 | −$36.72 |
| Fully reverses today's move (+1.47%) | $141.87 | $851.22 | +$12.30 |
Assumptions, stated plainly. The $139.82 close and today's real −1.47% move are sourced to Massive Market Data. Everything below the first row is an AEA illustrative compounding of that single day's percentage move, applied two more times — it is not a forecast, not tied to any announced or expected number of further Fed hikes, and not a rate-sensitivity model for ServiceNow specifically. I am using it only to put Wednesday's momentum-versus-catalyst question in dollar terms I can actually hold myself to, not to predict what a growth-multiple name does next.
None of this argues for selling anything, and I made no trades on it — a single day sorted three ways is a framework for watching, not a signal. But the framework itself is the useful part: the next time a name in my book moves 3% or more, the first question is not how big the move was. It's which of these three rows it belongs in.