thetrading.toolsSubscribe
SectorsDaily since 2022 · data as of November 20, 2026

AI Trade Index

The stocks traded as the AI build-out, in 9 groups along the chain: from the companies that pay for data centers to the ones that supply the chips, the equipment, the power and the software. One index for the whole trade, four price rules for each group, and a count of how many groups are in a bear market. Prices only.

Groups in a bear marketNovember 20, 2026
9 of 9
Every group
20% or more below their highest close of the past year
A week ago
7 of 9
+2 since Nov 13, 2026
AI Trade Index
602
−37.2% from its high
All four rules
3 of 9
groups meeting every rule
A count with a printed rule, never a score. The index is 100 on November 30, 2022.

Latest read

As of November 20, 2026, 9 of the 9 groups in the AI trade are 20% or more below their high of the past year. A week earlier the count was 7. 9 groups are under their 200-day average, 3 have a falling 200-day average and 9 are lagging the average stock. The AI Trade Index is at 602, 37.2% below its high of 958 on June 22, 2026. It started at 100 on November 30, 2022; the equal-weight S&P 500 is at 139 on the same scale. The deepest decline is in capacity lessors (70.2%) and the smallest in hyperscalers (21.2%).

Sources, methodology & freshnessDaily closing prices of the member stocks and of the equal-weight S&P 500 fund (RSP) · Every trading day after the close.Data as of 2026-11-20 · Open ↓
Source
Daily closing prices of the member stocks and of the equal-weight S&P 500 fund (RSP)
Methodology
Nine hand-picked groups of stocks, each an equal-weight index rebalanced daily; the AI Trade Index is the equal-weight average of the groups, 100 on November 30, 2022. Each group is read on four rules: 20% below its 252-session high, under its 200-day average, a 200-day average lower than 21 sessions earlier, and a ratio to the equal-weight S&P 500 under its own 200-day average. The headline counts groups in a bear market.
Updates
Every trading day after the close.Data as of 2026-11-20
Maintained & reviewed by Yuriy Matso — methodology shown on the page.
01

The AI Trade Index against the average stock

The index is the equal-weight average of the 9 groups below, set to 100 on November 30, 2022, the day ChatGPT was released. It reached 958 on June 22, 2026 and is at 602 as of November 20, 2026, 37.2% lower, after 107 sessions. That is 6.0 times its starting level. The equal-weight S&P 500, which counts every company the same and stands for the average stock, is at 1.4 times its level of the same day and 7.1% below its own high.

Range:

AI Trade Index

100 on November 30, 2022 · log scale · daily through November 20, 2026
1001k2023202420252026602783.5139
AI Trade Index200-day averageS&P 500, equal weight
Equal-weight index of 52 stocks in 9 groups, rebalanced daily, with its 200-day average and the equal-weight S&P 500 fund on the same scale. It cannot be bought as it is.
02

How many groups are in a bear market

A group is in a bear market when its index is 20% or more below its highest close of the past year. The count is 9 of 9 as of November 20, 2026. Every group counted has been in a bear market at once on one earlier stretch since 2022. From April 3, 2025 to April 23, 2025 all 8 groups then counted were, on 11 sessions. The index had fallen 33.6% from its high of January 22, 2025, made its low on April 4, 2025, and was at a new high 40 sessions later, on June 3, 2025. The average stock was as much as 18.5% below its own high at the time. The record is short, so this is history and not a forecast.

Range:

Groups 20% or more below their high

of 9 · daily through November 20, 2026
03692023202420252026AI Trade Index (top pane, log)6029 of 99
Groups in a bear marketGroups counted
The number of groups in a bear market on every session since November 30, 2022. The dashed line is the number of groups with enough history to be counted. The top pane is the AI Trade Index on a log scale.
03

Four price rules, group by group

Each rule is a common way of saying an uptrend has ended. As of November 20, 2026, 9 of 9 groups are in a bear market, 9 of 9 groups are under their 200-day average, 3 of 9 groups are with a falling 200-day average and 9 of 9 groups are lagging the average stock. 3 groups meet all four. Of the 52 stocks with enough history, 48 are under their own 200-day averages. The falling-average rule is the slowest: after a steep rise a 200-day average keeps climbing for months.

GroupFrom its highHigh on20% below its highUnder its 200-day average200-day average fallingLagging the average stockTo its 200-day average
Hyperscalers−21.2%Jun 1, 2026YesYesYesYes+11.4%
Capacity lessors−70.2%Jun 18, 2026YesYesYesYes+117.0%
Chip designers and their foundry−39.3%Oct 6, 2026YesYesNoYes+25.9%
Chip equipment−52.8%Jun 30, 2026YesYesNoYes+41.1%
Memory and storage−61.0%Jun 22, 2026YesYesNoYes+55.6%
Networking and optics−48.2%Jun 22, 2026YesYesNoYes+44.6%
Servers, cooling and electrical gear−37.7%Oct 6, 2026YesYesNoYes+23.2%
Power producers−32.9%Nov 28, 2025YesYesYesYes+22.8%
Software−27.3%Oct 26, 2026YesYesNoYes+7.2%
Groups meeting the rule9 of 99 of 93 of 99 of 9
The high is the highest close of the last 252 sessions. A 200-day average is falling when it is lower than 21 sessions earlier. The average stock is the equal-weight S&P 500: a group lags it when the group divided by that fund is under the ratio’s own 200-day average. The last column is the rise that would put the group back above its 200-day average as it stands today.

The groups did not peak together. In order of their highs: power producers (Nov 28, 2025), hyperscalers (Jun 1, 2026), capacity lessors (Jun 18, 2026), memory and storage (Jun 22, 2026), networking and optics (Jun 22, 2026), chip equipment (Jun 30, 2026), chip designers and their foundry (Oct 6, 2026), servers, cooling and electrical gear (Oct 6, 2026), software (Oct 26, 2026). The group nearest its 200-day average is software, 7.2% below it; the furthest is capacity lessors, which would need to rise 117%.

04

Hyperscalers

These companies own the largest data centers and pay for most of the build-out. The group’s index is at 268 as of November 20, 2026, 21.2% below its high of 340 on June 1, 2026, and −11.4% over the last five sessions. It is 2.7 times its level of November 30, 2022. It meets 4 of the four rules. 3 of its 5 stocks are under their own 200-day averages.

Range:

Hyperscalers: equal-weight index

100 on November 30, 2022 · log scale · through November 20, 2026
220202420252026AI Trade Index (top pane, log)602268298.8
Hyperscalers index200-day average
Equal-weight index of Microsoft, Alphabet, Amazon, Meta, Oracle, rebalanced daily, with its 200-day average. The top pane is the AI Trade Index on a log scale.
05

Capacity lessors

These companies build data centers and rent the computing power out. The group’s index is at 305 as of November 20, 2026, 70.2% below its high of 1023 on June 18, 2026, and −25.5% over the last five sessions. It is 3.0 times its level of March 28, 2025. It meets 4 of the four rules. 8 of its 8 stocks are under their own 200-day averages.

Range:

Capacity lessors: equal-weight index

100 on March 28, 2025 · log scale · through November 20, 2026
1001kAprJulOctJan 2026AprJulOctAI Trade Index (top pane, log)602305661.5
Capacity lessors index200-day average
Equal-weight index of CoreWeave, Nebius, IREN, Applied Digital, Core Scientific, Cipher Mining, TeraWulf, Hut 8, rebalanced daily, with its 200-day average. The top pane is the AI Trade Index on a log scale.
06

Chip designers and their foundry

These companies design and make the processors. The group’s index is at 566 as of November 20, 2026, 39.3% below its high of 932 on October 6, 2026, and −19.4% over the last five sessions. It is 5.7 times its level of November 30, 2022. It meets 3 of the four rules (not: 200-day average falling). 5 of its 5 stocks are under their own 200-day averages.

Range:

Chip designers and their foundry: equal-weight index

100 on November 30, 2022 · log scale · through November 20, 2026
360202420252026AI Trade Index (top pane, log)602566712.3
Chips index200-day average
Equal-weight index of Nvidia, AMD, Broadcom, Marvell, TSMC, rebalanced daily, with its 200-day average. The top pane is the AI Trade Index on a log scale.
07

Chip equipment

These companies make the machines that make the chips. The group’s index is at 306 as of November 20, 2026, 52.8% below its high of 648 on June 30, 2026, and −17.2% over the last five sessions. It is 3.1 times its level of November 30, 2022. It meets 3 of the four rules (not: 200-day average falling). 4 of its 4 stocks are under their own 200-day averages.

Range:

Chip equipment: equal-weight index

100 on November 30, 2022 · log scale · through November 20, 2026
260202420252026AI Trade Index (top pane, log)602306431.2
Chip equipment index200-day average
Equal-weight index of ASML, Applied Materials, Lam Research, KLA, rebalanced daily, with its 200-day average. The top pane is the AI Trade Index on a log scale.
08

Memory and storage

These companies make the memory chips and drives the servers need. The group’s index is at 1517 as of November 20, 2026, 61.0% below its high of 3895 on June 22, 2026, and −19.5% over the last five sessions. It is 15.2 times its level of November 30, 2022. It meets 3 of the four rules (not: 200-day average falling). 4 of its 4 stocks are under their own 200-day averages.

Range:

Memory and storage: equal-weight index

100 on November 30, 2022 · log scale · through November 20, 2026
1001k202420252026AI Trade Index (top pane, log)60215172360.6
Memory and storage index200-day average
Equal-weight index of Micron, SanDisk, Western Digital, Seagate, rebalanced daily, with its 200-day average. The top pane is the AI Trade Index on a log scale.
09

Networking and optics

These companies connect the servers to each other. The group’s index is at 838 as of November 20, 2026, 48.2% below its high of 1617 on June 22, 2026, and −18.6% over the last five sessions. It is 8.4 times its level of November 30, 2022. It meets 3 of the four rules (not: 200-day average falling). 6 of its 6 stocks are under their own 200-day averages.

Range:

Networking and optics: equal-weight index

100 on November 30, 2022 · log scale · through November 20, 2026
1001k202420252026AI Trade Index (top pane, log)6028381211.6
Networking index200-day average
Equal-weight index of Arista, Ciena, Coherent, Lumentum, Credo, Astera Labs, rebalanced daily, with its 200-day average. The top pane is the AI Trade Index on a log scale.
10

Servers, cooling and electrical gear

These companies build the servers and the equipment that powers and cools them. The group’s index is at 654 as of November 20, 2026, 37.7% below its high of 1050 on October 6, 2026, and −14.8% over the last five sessions. It is 6.5 times its level of November 30, 2022. It meets 3 of the four rules (not: 200-day average falling). 7 of its 7 stocks are under their own 200-day averages.

Range:

Servers, cooling and electrical gear: equal-weight index

100 on November 30, 2022 · log scale · through November 20, 2026
380202420252026AI Trade Index (top pane, log)602654806
Servers and gear index200-day average
Equal-weight index of Dell, Super Micro, HPE, Vertiv, Eaton, GE Vernova, Quanta Services, rebalanced daily, with its 200-day average. The top pane is the AI Trade Index on a log scale.
11

Power producers

These companies sell the electricity. The group’s index is at 395 as of November 20, 2026, 32.9% below its high of 589 on November 28, 2025, and −12.4% over the last five sessions. It is 4.0 times its level of November 30, 2022. It meets 4 of the four rules. 4 of its 4 stocks are under their own 200-day averages.

Range:

Power producers: equal-weight index

100 on November 30, 2022 · log scale · through November 20, 2026
250202420252026AI Trade Index (top pane, log)602395485.3
Power producers index200-day average
Equal-weight index of Vistra, Constellation, Talen, NRG, rebalanced daily, with its 200-day average. The top pane is the AI Trade Index on a log scale.
12

Software

These companies sell AI features to businesses, the payoff the spending is meant to earn. The group’s index is at 206 as of November 20, 2026, 27.3% below its high of 283 on October 26, 2026, and −14.3% over the last five sessions. It is 2.1 times its level of November 30, 2022. It meets 3 of the four rules (not: 200-day average falling). 7 of its 9 stocks are under their own 200-day averages.

Range:

Software: equal-weight index

100 on November 30, 2022 · log scale · through November 20, 2026
200202420252026AI Trade Index (top pane, log)602206220.5
Software index200-day average
Equal-weight index of Palantir, Salesforce, ServiceNow, Adobe, Snowflake, Datadog, MongoDB, Workday, Intuit, rebalanced daily, with its 200-day average. The top pane is the AI Trade Index on a log scale.
13

Limits of this index

  • The stocks are picked by hand as the companies most often traded as AI stocks. A different list would move the numbers. Changes to the list: 2026-11-22: first list, 52 scored stocks in nine groups, plus Anthropic described and not scored.
  • Every member is a company that is listed today, so any that failed along the way are missing and past levels read high.
  • The index is rebalanced every day and cannot be bought as it is. It measures the trade; it is not a product.
  • Groups need about a year of prices before the four rules can be read. Capacity lessors start on March 28, 2025, because before that most of the group were bitcoin miners.
  • AI model companies are listed and not scored: Anthropic has 2 sessions of prices, from November 19, 2026, and closed at 88.30.
  • The record is 4 years long. Nothing here is a base rate or a forecast, and the page does not say what to buy or sell.

How AI Trade Index Works

  1. 1
    Split the AI trade into groups
    The trade is a chain. Hyperscalers pay for data centers, capacity lessors rent computing power out, and chip, equipment, memory, networking, server and power companies supply them. Software sells the result to businesses. Each link is one group of stocks.
  2. 2
    Build one index per group, and one for the whole trade
    Inside a group every stock counts the same, so one very large company cannot decide the result. The nine groups, each counting the same, make the AI Trade Index. It starts at 100 on November 30, 2022, the day ChatGPT was released.
  3. 3
    Test each group on four price rules
    Is the group 20% or more below its high of the past year? Is it under its 200-day average? Is that average itself falling? Is the group lagging the equal-weight S&P 500, which stands for the average stock? Each rule is a common way of saying an uptrend has ended.
  4. 4
    Count the groups
    The headline is the number of groups in a bear market. It is a count with a printed rule, and every group keeps its own chart so the count can be checked.

Who Uses AI Trade Index

Traders in AI stocks
See which parts of the chain are still in an uptrend and which have lost it, without reading one stock at a time.
Index investors
Compare the AI trade with the average stock, to see whether a move in the index is the whole market or mostly this one theme.
Anyone following the build-out
Follow the order in which the groups peak and recover. The groups nearest the spending have not always moved with the chip makers.

Pro Tips

01
Read the dates of the highs, not only the depths
Groups in a chain rarely peak on the same day. Which group peaked first, and which held out longest, says more than the size of the fall.
02
The falling-average rule is the slow one
A 200-day average takes months to turn down after a steep rise. A group can be far below its high while its long average is still rising.
03
A full count is not a verdict
Every group counted has been in a bear market before, and the trade went on to new highs. The page lists each such stretch and what the index did next.

Common Issues & Solutions

The index cannot be bought▾
It is rebalanced every day so that each stock and each group counts the same. No fund does that. It is a measuring tool for the trade, not a product.
One group starts later than the others▾
The capacity lessors start when the first of them listed as an AI company. Before that most of the group were bitcoin miners, so their earlier prices say nothing about the AI trade.
A group is shown but not scored▾
A group needs about a year of prices before the four rules can be read. Until then it is listed with its members and left out of every count.
The list of stocks is a choice▾
The members are picked by hand as the companies most often traded as AI stocks. Every change to the list is dated on the page, because changing a member changes the history.

Frequently Asked Questions

What is the AI trade?▾
It is the group of stocks that investors buy and sell as a bet on the build-out of artificial intelligence. This page splits it into nine groups along the chain: hyperscalers, capacity lessors, chip designers and their foundry, chip equipment, memory and storage, networking and optics, servers and electrical gear, power producers, and software.
How is the AI Trade Index calculated?▾
Each group is an equal-weight index of its members' daily returns, rebalanced every day. The AI Trade Index is the equal-weight average of the nine groups. It is set to 100 on November 30, 2022, the day ChatGPT was released.
How do you decide whether the AI trade is in a bear market?▾
The page does not give one verdict. It counts how many of the nine groups are 20% or more below their highest close of the past year, and shows three more rules for each group: under its 200-day average, a falling 200-day average, and lagging the equal-weight S&P 500.
Why equal weight?▾
A few AI companies are among the largest in the world. Weighted by size, the index would mostly be those few stocks. Equal weight lets every company in a group, and every group in the trade, count the same.
Does a full count mean the AI trade is over?▾
No. The record shows an earlier stretch when every group counted was in a bear market, and the trade later made new highs. The page lists each such stretch with what the index did afterward. The record is short, so nothing on the page is a forecast.
Which stocks are in the index?▾
The members of every group are listed on the page with each stock's distance from its own high. They are chosen by hand as the companies most often traded as AI stocks, and every change to the list is dated.

Explore Other Tools