Is My Portfolio Exposed to AI Stocks (and an AI Bubble)?
If you own a US index or growth fund, you almost certainly own the AI trade, often more of it than you think. Here is how to measure it, how AI stocks behaved in past selloffs, and what the dot-com bust teaches.
Key takeaways
- Almost certainly, if you own a US index or growth fund. In Exent’s data, NVIDIA’s monthly returns had a 0.79 correlation with the Nasdaq-100 fund QQQ and 0.78 with a large-cap growth index fund over five years.
- Your AI exposure has two layers: stocks you hold directly and the same companies inside your funds. Add both before judging.
- AI-linked stocks have fallen hard before: NVIDIA, AMD and Meta each dropped more than 60% in the 2022 bear market, when the S&P 500 fell 25.4% (Exent’s data).
- The dot-com bust is the cautionary case: the Nasdaq Composite fell 77.9% from March 2000 to October 2002 and took until 2015 to regain its peak (FRED).
- “AI stocks” do not all move together. In 2026 through September, Micron rose 273.3% while Oracle fell 29.1%.
- Am I exposed to AI if I only own index funds?
- How do I measure my AI exposure?
- Which of my holdings move with AI stocks?
- How did AI stocks do in past selloffs?
- Is there an AI bubble, and what happens to my portfolio if it bursts?
- Do all AI stocks move together?
- What do investors do if their AI exposure is higher than intended?
Am I exposed to AI if I only own index funds?
Yes. The S&P 500 index and the Nasdaq‑100 weight companies by market value, and the biggest AI beneficiaries are among the largest companies in the world. The more an AI company’s value rises, the more of your index fund it becomes, without you doing anything.
That is not a flaw. It is how cap-weighted indexes work. But it means an investor with “just an index fund” plus a tech fund plus a few favorite stocks can hold the same handful of companies three times. See how concentrated the S&P 500 is and how much Magnificent 7 exposure you have.
How do I measure my AI exposure?
There are two complementary ways. The first counts what you own. The second measures how your holdings actually move.
- Look-through by weight. For each fund, take your weight in the fund and multiply it by the fund’s weight in each AI-linked company, from the fund’s holdings page. Add your direct stocks. This is called look-through analysis.
- Correlation. Check which of your holdings rise and fall together. Funds full of the same AI names move almost in lockstep, whatever their labels say.
An illustrative look-through for a three-holding portfolio:
| Holding (illustrative) | Your weight | AI-linked share inside (assumed) | AI-linked exposure |
|---|---|---|---|
| S&P 500 index fund | 50% | 30% | 15 pts |
| Tech sector fund | 20% | 60% | 12 pts |
| NVIDIA, held directly | 10% | 100% | 10 pts |
| Other stocks and bonds | 20% | 0% | 0 pts |
| Portfolio | 100% | 37% |
Illustrative only. The “AI-linked share” figures are assumptions for the example, not current fund weights. The point: a portfolio with 10% in one AI stock can have nearly four times that in AI exposure.
Which of my holdings move with AI stocks?
Broad growth funds and chip stocks track the AI leaders closely. Five-year correlations of monthly returns, from Exent’s data as of Sep 30, 2026 (1.0 means they move in lockstep):
| Holding | Correlation with QQQ | Correlation with S&P 500 (SPY) | Beta vs S&P 500 |
|---|---|---|---|
| NVIDIA (NVDA) | 0.79 | 0.69 | 2.21 |
| AMD (AMD) | 0.71 | 0.55 | 2.44 |
| Broadcom (AVGO) | 0.63 | 0.55 | 1.46 |
| Microsoft (MSFT) | 0.63 | 0.63 | 1.10 |
| Vertiv (VRT) | 0.51 | 0.50 | 2.03 |
| Semiconductor ETF (SMH) | 0.89 | 0.77 | 1.72 |
| Russell 1000 Growth ETF (IWF) | 0.98 | 0.94 | 1.16 |
NVIDIA’s most correlated large stocks in the same data were Amazon (0.64), Microsoft (0.61) and Synopsys (0.61). Its least correlated large caps included Bristol-Myers Squibb (−0.15) and Johnson & Johnson (−0.14). A beta above 2 means a stock has moved, on average, more than twice as much as the S&P 500. See how to check if your stocks are correlated.
How did AI stocks do in past selloffs?
They fell much more than the market, which is the other side of their big gains. Peak-to-trough falls in two recent episodes:
| Holding | 2022 inflation bear market | 2025 tariff selloff | Worst fall, last 5 years |
|---|---|---|---|
| NVIDIA | −61.8% | −32.7% | −66.3% |
| AMD | −61.6% | −31.9% | −65.5% |
| Meta Platforms | −62.3% | −28.3% | −76.7% |
| Micron | −49.8% | −38.0% | −57.6% |
| Vertiv | −66.3% | −44.9% | −71.3% |
| Semiconductor ETF (SMH) | −44.4% | −30.2% | −45.3% |
| Nasdaq-100 (QQQ) | −34.6% | −22.9% | −35.1% |
| S&P 500 (SPY) | −25.4% | −19.0% | −24.5% |
Source: Exent’s data, as of Sep 30, 2026. The same stocks then recovered and went on to large gains. A drawdown is only a permanent loss if you sell at the bottom, which is why position size matters more than the forecast. You can replay your own holdings through these episodes with the crash test.
Is there an AI bubble, and what happens to my portfolio if it bursts?
Nobody can confirm a bubble until after it bursts. What you can do is know how much of your money depends on one theme. The useful precedent is the dot-com era. The Nasdaq Composite closed at 5,048.62 on Mar 10, 2000, fell to 1,114.11 on Oct 9, 2002 (−77.9%), and did not close above its 2000 peak again until Apr 23, 2015 (FRED, price index).
The internet was real, and some survivors became the largest companies on earth. Investors still lost heavily, because being right about a technology is not the same as paying the right price for it. One difference today is that the largest AI companies earn large profits, which the leaders of 2000 often did not. That lowers the risk without removing it. For the valuation and concentration evidence, see are we in a stock market bubble?
Do all AI stocks move together?
Less and less. As AI spending meets the test of profits, the market has started to separate companies by results. Returns in 2026 through Sep 30, from the same data:
| Stock | 2026 YTD |
|---|---|
| Micron (MU) | +273.3% |
| AMD (AMD) | +185.7% |
| Arista Networks (ANET) | +55.4% |
| Vertiv (VRT) | +49.0% |
| NVIDIA (NVDA) | +22.5% |
| Microsoft (MSFT) | +6.5% |
| Broadcom (AVGO) | +1.7% |
| Constellation Energy (CEG) | −27.9% |
| Oracle (ORCL) | −29.1% |
A spread of more than 300 percentage points inside one theme means “how much AI do I own” is only half the question. The other half is which AI: companies already earning from it, or companies whose prices assume they will.
What do investors do if their AI exposure is higher than intended?
- Measure it first: direct stocks plus fund look-through plus correlation, as above.
- Size it for the drawdown, not the upside: ask what a 60% fall in that slice would cost in dollars, and whether you would hold.
- Avoid owning the same companies through several funds by accident; overlapping funds add concentration, not diversification.
- Add holdings that moved differently, not more of the same theme under another label.
- Set a rebalancing rule in advance, so a winning theme does not quietly take over the portfolio.
For tech specifically, see is my portfolio too concentrated in tech?
How to check your AI exposure with Exent
- Upload a screenshot of your holdings, or a CSV, Excel, PDF or Google Sheet; no brokerage login.
- Exent shows concentration and groups holdings that move together into correlation clusters, so AI‑linked stocks and growth funds show up as one bet.
- Crash replays show how the portfolio would have done in the 2022 and 2025 selloffs.
- Exent does not look inside ETF holdings, so the fund look‑through step is still yours to do.
Frequently asked questions
Add your direct AI-linked stocks to the AI share inside each fund you own, then compare the total with the loss you could tolerate if that slice fell 60%, as several AI leaders did in 2022.
Yes. The index weights companies by market value, and several of the largest are AI leaders. In Exent’s data, NVIDIA’s monthly returns had a 0.69 correlation with the S&P 500 ETF over five years.
Nobody knows if or when it would. As a reference, in the 2022 bear market NVIDIA fell 61.8% and the Nasdaq-100 fund QQQ 34.6% from peak to trough, and the Nasdaq Composite fell 77.9% after the 2000 dot-com peak.
There is no way to confirm a bubble in advance. Valuations and concentration are high by historical standards, but today’s largest AI companies are highly profitable, unlike many dot-com leaders.
Investors often start by removing overlap between funds, setting a maximum weight for the theme and directing new money elsewhere. This is a general approach, not a recommendation for your situation.
See what your portfolio is really made of.
Upload a screenshot of your holdings and get concentration, overlap, crash exposure and a risk-matched benchmark in about a minute.
Educational content, not investment advice. Examples are illustrative; past performance does not guarantee future results.