The AI Market Analyzer closed the week of October 30 at 31.8, in the Cooling zone. Five weeks ago it read 70.3 and was climbing toward AI Mania. The collapse came in two weeks of October, and the latest week barely moved, which tells you the selling has paused without any repair. The inputs have split in two. Everything measured from prices sits at or near the floor of the scale, while leveraged traders and company guidance still read as greedy.
Scoreboard
| Reading | Week of | Score | Zone | Change to now |
|---|---|---|---|---|
| Now | 2026-10-30 | 31.8 | Cooling | |
| Week ago | 2026-10-23 | 32.1 | Cooling | -0.3 |
| Month ago | 2026-09-25 | 70.3 | Hype | -38.5 |
| Year ago | 2025-10-24 | 85.4 | AI Mania | -53.6 |
Seven inputs, two stories
The score is the plain average of seven inputs, each scaled from 0 to 100 and computed from a 17-stock AI basket that spans chips, equipment, networking, power and the big cloud and software names. Four of the seven are in AI Winter. Two are in AI Mania. Only Volatility sits in between.
| Input | What it measures | Score | Zone | Latest reading |
|---|---|---|---|---|
| Momentum | Basket against its 25-week average | 8.2 | AI Winter | 6.2% below the 25-week average |
| Relative Strength | Basket 4-week return against the S&P 500 | 0.5 | AI Winter | 12.8 points behind the S&P 500 over 4 weeks |
| Breadth | Share of the 17 names above their 10-week average | 6.2 | AI Winter | 6%, about one name in seventeen |
| Volatility | 8-week volatility against its one-year norm, inverted | 33.2 | Cooling | 1.13 times the one-year norm |
| Speculation | Bull share of SOXL and SOXS trading | 87.0 | AI Mania | 69% bullish over 2 weeks |
| Rotation | Chips against software, 4-week change | 5.3 | AI Winter | Chips 13.9% behind software |
| Sentiment | Company guidance and press release tone | 82.2 | AI Mania | 16 guidance updates in 90 days |
Read together, the price inputs say the AI trade has lost almost all of its recent trend. Relative Strength at 0.5 is about as low as that input goes: the basket has trailed the broad market by close to 13 points in a month. Breadth at 6% means the weakness is not a story about one or two names. Nearly every stock in the basket is below its own 10-week average.
The two hot inputs point the other way, and for different reasons. Speculation at 87 shows that traders using the triple-leveraged chip funds kept buying the bull side through the drop. That is dip buying with borrowed exposure, and it is the opposite of what a washed-out bottom normally looks like. Sentiment at 82.2 reflects what the companies themselves are saying, and their guidance through the third-quarter reporting season stayed upbeat. Volatility is elevated but not extreme at 1.13 times its norm, which fits a steady grind lower more than a panic.
What moved the index
The index fell 38.4 points across the weeks ending October 9 and October 16, the third-largest two-week drop in its three-year history. The trigger was a cluster of doubts about how the AI build-out gets paid for. Broadcom’s chief executive said access to capital, more than chip supply, could become the main limit on AI expansion. The IMF warned that weaker returns on AI investment could set off a correction in technology shares. A hotter September inflation print, at 3.7% a year, then raised the odds of higher rates at exactly the moment the spending plans depend on cheap financing. Chip stocks put together a seven-session losing streak and the Philadelphia Semiconductor Index lost about a tenth of its value in six sessions.
The second half of the month was earnings season, and it kept the index pinned down. Meta’s capital spending ran ahead of its operating cash flow, and Alphabet fell after reporting negative free cash flow for the quarter. Both stories are covered in the AI Capex Tracker for November 2026, which shows the big four spending $237 billion in a single quarter. Through all of it, money moved from chips toward software, which is why Rotation sits at 5.3. The market is still paying for AI. It has stopped paying a premium for the companies that sell the hardware.
Three years of readings
Out of 157 weekly readings, only 16, about one in ten, have been at or below today’s 31.8. The index has spent the most time in Hype (61 weeks) and the least in AI Winter (8 weeks). The all-time high of 89.2 came in February 2024. The low of 13.0 came in April 2025, when a broad market selloff hit the AI names hardest. Averages by year sit in a narrow band, 56.4 in 2024, 55.8 in 2025 and 58.6 so far in 2026, so the index tends to swing hard around a slightly warm middle.
The previous two-week drops of a similar size did not resolve the same way, which is a useful warning against treating this reading as a timing signal.
| Two-week drop | Size | Reading | 4 weeks later | 13 weeks later |
|---|---|---|---|---|
| April 2024 | -36.4 | 32.7 | 67.4 | 24.2 |
| July 2024 | -44.2 | 24.2 | 35.5 | 47.3 |
| February 2025 | -37.5 | 39.3 | 29.5 | 43.8 |
| October 2026 | -38.4 | 36.8 | to come | to come |
In April 2024 the index was back in Hype within a month. In July 2024 it stayed in Cooling or lower for most of the following quarter. In February 2025 the first drop was followed by a deeper one, down to the record low. The current episode has already lasted three weeks below 37.
What to watch
- Breadth. A move back above roughly a third of the basket trading over its 10-week average would be the first sign that buying is spreading again.
- Speculation. If leveraged bull positioning gives way, the input would drop sharply and could pull the score into AI Winter. A washout of that kind is what a durable low usually looks like, and it has not happened yet.
- Rotation. Chips have trailed software for a month. A turn here would say investors are again willing to pay up for suppliers.
- Nvidia’s quarterly report later in November, the next inflation print and the December Fed decision are the scheduled events most likely to move the price inputs.
- Financing news. Bond deals, private credit and lease structures for data centers now move the AI trade as much as product news does.
The live AI Market Analyzer updates through the week, so check it for the reading since this post.
Bottom line
The AI trade has gone from near Mania to Cooling in five weeks, and the damage is broad: momentum, relative strength, breadth and the chips-versus-software spread are all at levels the index has rarely seen. What has not happened yet is a flush of the leveraged bulls, and company guidance remains confident, so prices have cooled faster than conviction. For decision makers this argues for patience over action in either direction. The fundamental spending story is intact, but the market is now demanding proof that the spending earns a return and that it can be financed, and past episodes of this size have ended in a fast rebound as often as in a deeper low. Watch breadth and leveraged positioning for the turn rather than trying to call it from the headline score.
Method. The AI Market Analyzer averages seven equal-weight inputs, each scaled from 0 to 100, computed weekly from a 17-stock AI basket (NVDA, AVGO, AMD, TSM, MU, ASML, MRVL, AMAT, LRCX, ANET, VRT, MSFT, GOOGL, META, AMZN, ORCL, PLTR) plus leveraged chip ETF trading, chip and software ETFs, and company guidance and press release tone. Zones: 0 to 24 AI Winter, 25 to 44 Cooling, 45 to 55 Neutral, 56 to 75 Hype, 76 to 100 AI Mania. Readings here are as of the week ending 2026-10-30. This is a market sentiment indicator, not investment advice.
