Stress Testing Amid Rising Fears of an AI Bubble

In the light of recent market reactions to rising fears of an artificial intelligence bubble, we have conducted stress testing on investable indices and strategies as a response to a hypothetical AI bubble burst. As concerns of tech companies being overpriced is often compared to the dot-com bubble burst from 2000, we will use this historical period and respective responses of market factors as a starting point for our analysis.

Dot-Com Bubble Burst

During the era of the dot-com bubble, the market value of tech companies peaked in March 2000 before the bubble burst at the end of the month. They subsequently lost value on the market steadily and continuously for more than two years.

We explore the responses of a few investable indices below during this period, with both a short-term focus of a 2-month period and a long-term focus of 2.5 years. While we plot the following S&P Index for context, we focus more on the indices with an information technology profile where, as expected for a tech company, the shock response is more acute. We also explore the reaction of the S&P Information Technology sector index, as well as the NASDAQ Composite Index and NASDAQ-100 Index.

Compared to the broader S&P 500 index, the S&P Information Technology sector loses more value and at a faster pace due to its focus on technology companies. Similarly, the NASDAQ 100 drops faster and loses more than the NASDAQ Composite index. And while both have significant concentration in technology companies, the NASDAQ 100 holds large companies that suffered more from this historical market turbulence.

Therefore, to recreate a similar bubble burst for AI companies, we shocked the value of the NASDAQ 100 and S&P Information Technology sector to account for the specifics of the market crash and its impact in more detail.

The table below highlights the exact percentage returns of the indices during the short- and long-term periods we explored.

Technical and AI-Focused Companies Sell-Off from November 2025

Now we turn back to recent days and analyze the reactions of the same indices during the last couple of weeks, when AI- and technology- related companies began losing market value based on sell-offs of their stocks. A hint of a possible AI bubble burst—vocalized by IMF and Bank of England representatives—was also visible in the market response and returns of investable indices holding the AI-related stocks. Returns are presented in the table below.

The main difference in the parallel drop of the indices compared to the 2000 dot-com bubble burst is that the S&P Information Technology index drop was slightly larger than NASDAQ indices’ drop, based on its constituent concentration today. So, when we define the AI bubble burst stress testing scenarios in next section, we account for the more recent response of the S&P Information Technology index and have set similar shock magnitudes for this index and the NASDAQ 100 index.

AI Bubble Burst Hypothetical Scenarios

We next define three scenarios to account for how acute the hypothetical impact on market can be from an AI-related sell-off or a short- and long-term bubble burst.

• AI sell-off - Starting with the lowest-impact scenario, we define the AI sell-off from the last couple of weeks with a 4%-5% drop in technical-focused equity indices and a 1.5% drop in bond indices.

• AI bubble burst short term – This scenario replicates the first 2 months after the hypothetical AI bubble burst with a 35% drop in technical-focused equity indices and a 3% drop in bond indices.

• AI bubble burst long term - This scenario replicates the period of 2.5 years after the hypothetical AI bubble burst with an 80 % drop in technical equity indices and a 3% drop in bond indices.

Scenario Analysis of AI Sell-Off and the Hypothetical AI Bubble Burst

We analyze the impact from the above scenarios on a few standard investable strategies with different allocations in equity and bond indices. The strategies below are ordered by their allocation in equities in descending order:

• Aggressive Growth – 100% equities

• Growth – 80% equities, 20% corporate bonds

• Moderate – 60% equities, 40% corporate bonds

• Balanced – 50% equities, 50% corporate bonds

• Conservative – 40% equities, 40% corporate bonds, 20% government bonds

• Income – 20% equities, 40% corporate bonds, 40% government bonds

• Core Bond – 50% corporate bonds, 50% government bonds

As expected, the strategies with higher allocations to equities would have higher impact from the hypothetical AI scenarios, which is in line with the impact observed during the dot-com bubble burst shock: A sharp drop in equity indices with a very slight drop in bond indices.

These market shocks are driven by equity markets and, in particular, the technical-related stocks. Therefore, the impact on more diversified equity indices is lower than technology-focused indices' negative returns. With investment strategy mixes that include bonds, the impact is even lower.

Conclusion

The above analysis is an example of hypothetical stress testing for an AI-related bubble burst, tracking closely to the historically observed market impact of the dot-com bubble burst in 2000. The above approach will be implemented via new Thematic Scenarios to be added to the suite of template scenarios in the FactSet Workstation in the coming weeks. Clients may use them as a base to add their views and speculations on possible impacts to market indices.

This blog post is for informational purposes only. The information contained in this blog post is not legal, tax, or investment advice. FactSet does not endorse or recommend any investments and assumes no liability for any consequence relating directly or indirectly to any action or inaction taken based on the information contained in this article.