Which AI layer is cracking first? Week 1

Our read: in this AI selloff, software is the only layer holding up; in every other layer most or all names are below their 50-day.

Software is the only group where every name we track is back above its 50-day average (up from 57% a week earlier, our calc).

Quick key: "above the 200-day" means a stock is trading above its average price of the last ~200 trading days, a rough test of whether its long-term trend is still up. The 50-day is the same idea over a shorter window. RSI is a 0–100 momentum score; below 30 is unusually weak, above 70 unusually strong.

Share of names above the 50-day / above the 200-day / average RSI:

• Software: 100% / 100% / 57.7
• Big spenders: 20% / 40% / 40.8
• Semis: 0% / 50% / 37.7
• Optical: 0% / 43% / 36.7
• Power: 0% / 0% / 37.2
• Servers & storage: 0% / 100% / 36.8
• Neoclouds: 0% / 0% / 32.1 (weakest)

The other big move: big spenders above the 200-day fell from 80% to 40%. Note that $META is right at its 200-day (0.16% under), and $MDB and $ASML are also hairline-close, so small moves can flip these counts.

What would change the read

Set before publication. We report back in public when one trips.

• if neoclouds and power get names back above the 50-day while software slips below it, the weakness is rotating, not concentrated in hardware. Open · next check: Weekly update (next: Nov 2 close)

We'll update this every week.

General research, the same for every reader. Not investment advice.

Sources: our calc from daily closes, 10/26 vs 10/19.

Method and sources

Figures marked "our calc" are our own arithmetic on the cited primary sources (company filings, earnings releases, public datasets). Every figure was checked against its source before publication.

Original post on X → Download data note

General research, the same for every reader. Not investment advice.