ResourcesGuide

What Happens to Your Portfolio if the AI Bubble Bursts?

Nobody knows if AI stocks are a bubble. What history shows is how far tech, chip stocks, Nvidia and leveraged funds fell in the dot‑com bust and in every AI‑era selloff since 2021, what held up, and how long recovery took. With a market risk check updated every trading day.

Published Updated 8 min readHow we research

Key takeaways

  • The full‑size precedent is the dot‑com bust: the Nasdaq‑100 fund QQQ fell 83% from March 2000 to October 2002 and took about 15 years to break even with dividends.
  • The AI‑era selloffs since 2021 were smaller but followed the same order: the S&P 500 fell least, the Nasdaq‑100 more, chip stocks and Nvidia more again, and leveraged chip funds most. The 3× chip fund SOXL lost 61.8% in 20 trading days in summer 2024 (Exent’s data).
  • One bad day can do a lot: on Jan 27, 2025 Nvidia fell 17%, the largest one‑day loss of market value in US history, while the S&P 500 fund fell 1.4%.
  • T‑bills gained in every selloff. Long Treasuries rose in the 2024 and 2025 chip selloffs but fell 34.0% in the 2022 tech bear market, when rates were rising.
  • Recovery time grew with concentration and leverage: QQQ needed about two years after its 2021 record, SOXL 18 months after summer 2024.

Market risk reading, cross‑asset context and an events calendar: see Exent →

What happens to a portfolio if the AI bubble bursts?

Nobody can say in advance whether AI stocks are a bubble; that is only clear afterwards. What can be measured is what happened to portfolios in past tech busts. The largest was the dot‑com bust: QQQ, which tracks the Nasdaq‑100, closed at $117.88 on Mar 24, 2000 and at $20.06 on Oct 9, 2002, a fall of 83%. It did not close above its 2000 peak again for more than 16 years, and an investor who reinvested dividends waited about 15 years to break even (The Motley Fool on Yahoo Finance).

Since 2021 there have been three large tech and chip selloffs, plus one shock day. The 2022 fall came before the AI boom, in a year of fast rate hikes; the 2024 and 2025 chip selloffs came in the middle of it. Each window below runs from a record close to the low that followed, dividends included. This page is part of the portfolio protection guide; for whether today’s prices look stretched, see are we in a stock market bubble.

EpisodeWindow (close to close)Trading daysS&P 500 (SPY), total returnS&P 500’s largest fall inside (price)
2022 tech bear marketNov 19, 2021 → Nov 3, 2022240−19.7%−25.4%
Summer 2024 chip selloffJul 10, 2024 → Aug 7, 202420−7.6%−8.4%
2025 chip selloffJan 22, 2025 → Apr 8, 202553−17.9%−19.0%
Episode2022 tech bear marketSummer 2024 chip selloff2025 chip selloff
Window (close to close)Nov 19, 2021 → Nov 3, 2022Jul 10, 2024 → Aug 7, 2024Jan 22, 2025 → Apr 8, 2025
Trading days2402053
S&P 500 (SPY), total return−19.7%−7.6%−17.9%
S&P 500’s largest fall inside (price)−25.4%−8.4%−19.0%

Market risk check, updated every trading day

Data through the close of Nov 2, 2026. The S&P 500 fell 1.7% over the last five trading days, and the Nasdaq‑100 fell 4.4%. The S&P 500 is 3.5% below its record close of Oct 6: a small dip from its record. It is 3.1% above its 200‑day average (+5.2% a week earlier). Its 20‑day volatility is 11.3% a year, about normal for the period since 2016.

The S&P 500 vs its 200‑day average

Daily closes over the last year, through Nov 2, 2026. Green: the index above its average; red: below it.

−3.5%from its record closeRecord: Oct 6, 2026
+3.1%vs its 200‑day averageA week earlier: +5.2%
11.3%20‑day volatility (a year)Median since 2016: 12.2%
−1.7%last five trading daysNasdaq‑100: −4.4%

Leveraged funds this year: the fund vs its multiple

Year‑to‑date return through Nov 2, 2026: what each fund returned next to its multiple times the index’s return. The gap is the daily reset at work.

SSO
UPRO
QLD
TQQQ
TNA
FundLast 5 daysYear to dateMultiple × indexFrom its high
2× S&P 500 (SSO)−3.5%+16.4%+22.3%−7.4%
3× S&P 500 (UPRO)−5.4%+20.9%+33.5%−11.6%
2× Nasdaq‑100 (QLD)−8.8%+14.2%+22.5%−20.3%
3× Nasdaq‑100 (TQQQ)−13.1%+13.5%+33.7%−31.5%
3× Russell 2000 (TNA)+2.1%+26.7%+42.5%−46.4%
3× chip stocks (SOXL)−24.7%+62.3%+152.7%−77.3%
  • 2× S&P 500 (SSO)
    Last 5 days
    −3.5%
    Year to date
    +16.4%
    Multiple × index
    +22.3%
    From its high
    −7.4%
  • 3× S&P 500 (UPRO)
    Last 5 days
    −5.4%
    Year to date
    +20.9%
    Multiple × index
    +33.5%
    From its high
    −11.6%
  • 2× Nasdaq‑100 (QLD)
    Last 5 days
    −8.8%
    Year to date
    +14.2%
    Multiple × index
    +22.5%
    From its high
    −20.3%
  • 3× Nasdaq‑100 (TQQQ)
    Last 5 days
    −13.1%
    Year to date
    +13.5%
    Multiple × index
    +33.7%
    From its high
    −31.5%
  • 3× Russell 2000 (TNA)
    Last 5 days
    +2.1%
    Year to date
    +26.7%
    Multiple × index
    +42.5%
    From its high
    −46.4%
  • 3× chip stocks (SOXL)
    Last 5 days
    −24.7%
    Year to date
    +62.3%
    Multiple × index
    +152.7%
    From its high
    −77.3%

“Multiple × index” = 2× or 3× the index fund’s return since Dec 31, 2025; a fund resets its leverage daily, so it never promises that multiple over longer periods.

Exent macro regime score, today

Last updated Nov 2, 2026. Members only.

Exent regime pillarToday
Liquidity
Credit stress
Economic cycle
Sentiment
Regime: Risk ON · Neutral · Risk OFF

Price trend and volatility show what the market has already done. Exent’s macro regime score reads the backdrop behind it (liquidity, credit stress, the economic cycle and sentiment) and classifies it as Risk ON, Neutral or Risk OFF. Market intel members see today’s reading and get an alert when it changes.

Exent market data, daily closes; returns include dividends and interest. The S&P 500, Nasdaq‑100 and Russell 2000 are shown through the SPY, QQQ and IWM funds. Refreshed every trading day. It describes conditions; it is not a forecast or a recommendation. Past performance does not guarantee future results.

How far did AI stocks, chip stocks and leveraged funds fall?

From each window’s start to its low, the S&P 500 fell less than the Nasdaq‑100, chip stocks and Nvidia fell more, and the 3× chip fund fell most:

Fund2022 tech bearSummer 2024Early 2025
S&P 500 (SPY)−19.7%−7.6%−17.9%
Min‑volatility US stocks (USMV)−9.9%+1.0%−5.4%
Nasdaq‑100 (QQQ)−35.1%−13.6%−21.6%
Chip stocks (SOXX)−40.9%−24.9%−34.2%
Nvidia (NVDA)−59.3%−26.7%−34.5%
2× Nasdaq‑100 (QLD)−62.9%−26.2%−40.4%
3× Nasdaq‑100 (TQQQ)−80.6%−37.4%−55.4%
3× chip stocks (SOXL)−88.4%−61.8%−76.5%

The broad S&P 500 lost 7.6% to 19.7%, the Nasdaq‑100 13.6% to 35.1%, chip stocks 24.9% to 40.9%. Leverage on top of a concentrated theme was the most dangerous combination: the 3× chip fund SOXL lost 61.8% in the 20 trading days of the summer 2024 selloff and 88.4% over the 2022 tech bear market. The 3× Nasdaq‑100 fund TQQQ lost 80.6% in 2022. A fund that loses 88% needs to rise more than eightfold to get back to even (loss recovery math). More on those funds in how much leveraged ETFs lose in a crash and how much QQQ can fall.

What one bad AI day looked like: Jan 27, 2025

When a Chinese start‑up, DeepSeek, showed a low‑cost AI model, Nvidia fell 17% in a day and lost $589 billion of market value, the largest one‑day loss in US stock‑market history (Bloomberg on Yahoo Finance). The broad market fell far less, which is what concentration risk looks like:

FundFirst dayFirst 5 trading days
Nvidia (NVDA)−17.0%−15.8%
Chip stocks (SOXX)−7.8%−5.4%
3× chip stocks (SOXL)−23.2%−17.3%
Nasdaq‑100 (QQQ)−2.9%−1.4%
3× Nasdaq‑100 (TQQQ)−8.8%−4.7%
S&P 500 (SPY)−1.4%−1.0%
Min‑volatility US stocks (USMV)+1.1%+1.1%
Long Treasuries (TLT)+1.2%+0.6%

A portfolio that held the S&P 500 fund lost 1.4% that day; one that held Nvidia lost 17%, and one that held the 3× chip fund lost 23.2%. Stock‑level detail on how much AI a portfolio really holds is in how much AI exposure is in your portfolio.

What held up when tech stocks fell?

Fund2022 tech bearSummer 2024Early 2025
Long Treasuries (TLT)−34.0%+3.9%+2.0%
7–10y Treasuries (IEF)−17.2%+2.9%+3.6%
Gold (GLD)−12.1%+0.5%+8.2%
T‑bills (BIL)+0.8%+0.4%+0.9%
Energy stocks (XLE)+72.7%−2.2%−16.1%

T‑bills (BIL) gained in all three windows. Treasuries were mixed: they rose in the 2024 and 2025 chip selloffs, and lost 34.0% (long) and 17.2% (7–10 year) in the 2022 window, a year in which the Fed raised rates seven times. Energy stocks rose 72.7% in the 2022 window as oil climbed, a reminder that a tech bust is not a whole‑market bust. Minimum‑ volatility stocks (USMV) fell far less than the Nasdaq‑100 every time. Where to wait out a fall is compared in cash, T‑bills or bonds when you step aside.

How long did recovery take?

When each fund first closed back at its value from the start of each window (data through Sep 30, 2026):

Back to its starting value2022 tech bearSummer 2024Early 2025
S&P 500 (SPY)Jul 2023 (20 mo.)Aug 2024 (1 mo.)Jun 2025 (5 mo.)
Nasdaq‑100 (QQQ)Dec 2023 (25 mo.)Nov 2024 (4 mo.)Jun 2025 (5 mo.)
Chip stocks (SOXX)Jul 2023 (20 mo.)Sep 2025 (14 mo.)Jun 2025 (5 mo.)
Nvidia (NVDA)May 2023 (18 mo.)Oct 2024 (3 mo.)Jun 2025 (5 mo.)
3× Nasdaq‑100 (TQQQ)Jul 2024 (32 mo.)Dec 2024 (5 mo.)Jul 2025 (6 mo.)
3× chip stocks (SOXL)Jul 2024 (32 mo.)Jan 2026 (19 mo.)Oct 2025 (8 mo.)

The recent selloffs recovered within months to two years. The dot‑com bust shows the other outcome: the index needed about 15 years. Which outcome a future AI selloff would resemble is not knowable in advance; how much of a portfolio rides on it is.

How investors limit AI‑bubble risk

  • Measuring concentration first: how much of the portfolio sits in a handful of AI and chip names, index funds included; see do I have too much tech exposure.
  • Keeping leverage off the most concentrated themes. In every window above, the 3× chip fund lost two to three times as much as the chip index.
  • Writing an exit rule before the selloff, for example a trend rule; see TQQQ and the 200‑day moving average (a hypothetical backtest).
  • Knowing when high‑beta exposure has historically paid and when it has not: when to own high‑beta stocks.

How Exent helps you manage a concentrated bet

  1. Exent Predictive Market Intelligence publishes a daily macro regime score (Risk ON, Neutral or Risk OFF) from four pillars: liquidity, credit stress, the economic cycle and sentiment, with an alert when it changes.
  2. Its six model allocations include a 2× model that uses leverage only in Risk ON and steps aside to cash in Risk OFF.
  3. Portfolio analysis shows how much of your portfolio rides on AI and chip stocks and replays your holdings through past crashes, with the loss in dollars; upload a screenshot, CSV, Excel, PDF or Google Sheet, no brokerage login.

Featured on Business Insider · Yahoo Finance · Benzinga

Exent is our product. Educational content, not investment advice.

Frequently asked questions

Nobody knows. In the dot-com bust the Nasdaq-100 fund QQQ fell 83% from March 2000 to October 2002. In the 2022 tech bear market QQQ fell 35.1%, the S&P 500 fund 19.7% and the 3x chip fund SOXL 88.4%, in Exent’s data.

It fell 59.3% from November 2021 to November 2022 (66.3% at its worst), 26.7% in the summer 2024 chip selloff, 34.5% from January to April 2025, and 17% on January 27, 2025 alone.

They fall several times as far as the index. In the 2022 tech bear market TQQQ lost 80.6% and SOXL 88.4%; in the 20 trading days of the summer 2024 chip selloff SOXL lost 61.8%.

In the three tech selloffs since 2021, T-bills gained every time and minimum-volatility stocks fell far less than the Nasdaq-100. Treasuries rose in the 2024 and 2025 chip selloffs but fell sharply in 2022, when rates were rising.

QQQ got back to its November 2021 value in December 2023. After the summer 2024 selloff the chip fund SOXX needed until September 2025 and SOXL until January 2026. After the dot-com bust QQQ took about 15 years with dividends.

Add market context to your plan.

Market intel provides a Risk ON / Neutral / Risk OFF reading from four pillars, cross-asset analysis, an events calendar and regular written notes. Context for reviewing a plan, not a crash forecast.

See what Market intel includesLimited offer67% off your first month

Educational content, not investment advice. Examples are illustrative; past performance does not guarantee future results.

DISCLAIMER: Exent AI is an informational research platform and does not provide investment advice, financial advice, or act as an investment adviser, broker-dealer, or fiduciary. All content, analytics, and model outputs are provided for educational and informational purposes only and should not be relied upon as a recommendation to buy, sell, or hold any financial instrument. Investing involves risk, including the possible loss of principal. Exent AI does not execute trades, hold customer funds, or provide brokerage services. Past performance and hypothetical results are not indicative of future results. Third-party data and AI-generated outputs may be incomplete or inaccurate.