Is today’s U.S. stock market riskier than the top of the internet bubble in 2000? While 2025 bulls point to a lack of a frothy IPO market, 2025 bears point to anything with AI in the description trading at obscene valuations, along with quantum computing, space travel, robots, flying cars, and other unprofitable investments. But what if the answer is “both”, meaning a high level of risk is tied to a handful of the largest companies, while hundreds of companies trade at discounted valuations with inflecting fundamentals?
S&P 500 valuation: Presently, the S&P 500 trades at a price-to-earnings ratio of over 22x. To put this in a historic context, since the internet bubble burst 25 years ago, the S&P 500 has traded between 9x (March 2009) and 26x (top of the internet bubble). The long-term average is 16.2x, equating to the index trading 38% above the long-term average today. Statistically, today’s S&P 500 valuation is two-standard deviations above the average (orange horizontal line), meaning valuation should only be this high about 2.5% of the time. Initial conclusion – the S&P 500 is overvalued.
Source: FactSet and Osborne Partners
However, researching the composition of the S&P 500 yields a different take. Today, the S&P 500 is far more concentrated than any time in history. In fact, the 8 largest holdings in the entire index represent over 37% of the index. As a comparison, at the top of the internet bubble in 2000, the 10 largest companies were 26% of the index.
However, concentration risk is only one of six major risks for this cohort of 8 companies.
The risks include:
- Concentration
- Valuation
- Interdependence
- Customer exposure
- Competition
- Correlations to each other
Besides the S&P 500 being 37% concentrated in 8 companies, the valuation differential between the 8 largest and the other 492 is severe. As the following table shows, while the S&P 500 trades at 22.3x (2026 earnings), the 8 largest (37% of the index), trade at a P/E on 2026 earnings of 38x. So while the S&P 500 trades at a 38% premium to the long-term average, the 8 largest companies trade at a 72% premium to the “overvalued” S&P 500.
Source: FactSet and Osborne Partners
Meanwhile, the approximate total value of the other 492 companies divided by the estimated earnings of the 492 points to a P/E of well under 19x, making the 8 largest 100% more expensive than the 492.
However, as the internet bubble proved, valuations can remain overinflated for long periods of time. So while concentration and valuation risks can continue, the fundamental nature of these companies are where the more serious risk may reside.
What makes today potentially riskier are the interdependence, customer exposure, competition, and correlations of these 8 that are 37% of the index. Many investors are unaware of these risks.
Interdependence: Many of these 8 largest are dependent on each other. Here are three examples:
- NVDA – Data center revenue is 88% of the total company revenue. Depending on the quarter, we estimate 55-65% of revenue is derived from the other 7 largest companies. For example, in the most recent two quarters in 2025, approximately 50% of NVDA’s growth was derived from MSFT and AMZN alone.
- AMZN – Hosts both META and AAPL data in their datacenters.
- AVGO – Depending on the quarter, we estimate approximately 56-60% of total company revenue comes from selling cloud software and semiconductors, along with semiconductors for AI (custom AI accelerator chips and networking chips for AI related datacenters).
In contrast, at the top of the internet bubble, the 10 largest companies were far less dependent on each other and far more diverse.
Source: S&P 500
Customer and end-market exposure: While all of these companies are scrambling to buy as many semiconductors as possible from NVDA, they are using the semis to serve many of the same customer types in the same sectors.
Exposure to cloud storage:
- MSFT, GOOGL, and AMZN are 70% of the total market.
- MSFT cloud revenue is 38% of total revenue.
- AMZN cloud revenue is 20% of total revenue.
- GOOGL cloud revenue is 17% of total revenue.
- AVGO semis and software revenue for the cloud are 25% of total revenue.
Exposure to AI cloud based and consumer based paints the same picture:
- Cloud corporate AI largest companies = MSFT, GOOGL, AMZN, META
- Consumer AI largest companies = AAPL, META, AMZN
Exposure to online ad market. Three of the 8 largest companies total 60% of the online ad market.
- GOOGL, META, and AMZN are approximately 60% of total market
Competition: The interdependence and end market exposure means most of the 8 largest are competing for the same customers:
- Semiconductors: 88% of NVDA’s revenue mostly builds out data centers. 55-65% of revenue is from 6 customers who directly compete with each other in the same sectors. Additionally, NVDA competes with AMD, INTC, QCOM, AVGO, and in-house created chips from GOOGL (Tensor), AMZN (Trainium), and MSFT (Maia).
- Cloud storage: MSFT, GOOGL, AMZN plus BABA and ORCL, we estimate to be about 80% of the global storage market. This is a cut throat end market with high capex.
- AI: The largest 7, excluding NVDA, derived less than 10% of revenue from AI, but the spend is massive for MSFT, AAPL, AMZN, and GOOGL.
- Online ads – As previously shown, three of the largest seven are 60% of the market.
Correlations: The result of the interdependence, customer exposure, and competition is that the stocks of these companies are increasingly moving in tandem. The following table shows the correlations between each other for the past five quarters (3Q 2024 to today).
- 1.0 = Perfect positive correlation.
- 0.0 = No correlation.
- -1.0 = Perfect negative correlation.
Example: A correlation of 0.80 means if one rises/falls 10%, the other rises/falls 8%.
So not only is 37% of the S&P 500 comprised of these 8 companies, but these 8 companies trade at a 38x P/E on 2026 earnings, while being interdependent on each other, having exposure to the same end markets and customers, and competing against each other, while their stock prices move in the same direction by as much as a factor of 0.88.
At Osborne Partners, while we have exposure to some of the largest 8, we are careful about correlations, valuation, and position sizes within portfolios.
But what about the other 63% of the S&P 500 in 492 companies?
These companies trade at less than half of the valuation of the largest 8. However, a large swath of the 492 receive fundamental tailwinds from a lower Fed Funds target rate, which is now 4.25% from a recent peak of 5.50%. Examples can be found in the Consumer Discretionary and Financial sectors. Additionally, there are sectors that are suddenly trading near record discounts to the S&P 500 due to recent underperformance. One example is the healthcare sector that now trades at a 26% discount to the S&P 500 versus a 0% discount just a few years ago.
Source: FactSet
We are delivering a dual message here. First, although the 8 largest are high quality companies, investors should be careful about high exposure to them due to the fact they mirror each other in many ways. Second, if and when these 8 return to some version of normalcy via a rerating lower, there is plenty of value and positively inflecting fundamentals in the other 492 companies in the S&P 500. Just as we felt many of the largest 8 were generational values exiting the global financial crisis in 2009, we now believe many of the best future opportunities lie outside these 8.
