The S&P 500 has been setting records. The headline number looks clean. But the stocks driving those records are a narrow group, and most of the index is not going along for the ride.
That split is not unusual in a late-cycle market. What makes this one notable is the degree. The gap between a handful of AI-driven names and the rest of the market has grown to a point that, by one measure, has only one historical parallel.
Michael Burry made his name betting against the U.S. housing market before the 2008 crisis. He has spent recent weeks studying four decades of S&P 500 data, comparing what he sees now to every major market top since 1985. He keeps arriving at the same year: March 2000.
Michael Burry says this market looks a lot like March 2000
Burry tracks what he calls the Bear Market Ratio: the share of S&P 500 members trading more than 20% below their 10-year highs. When the index closed at a record on Oct. 6, that reading stood at 55%. The only time in four decades of data it came in worse was at the March 2000 peak. Typical market tops over that same period ran between roughly 20% and 31%.
Also read: Michael Burry just offered young Americans a reality check on housing
Breadth tells the same story. On the day of that October record, just 15 S&P 500 stocks closed at a new high alongside the index. That number was identical to what Burry found in March 2000, when the index was also setting records while most of its members were quietly declining. The S&P 500 now trades at a median price about 22% below its 10-year high.
The Magnificent Seven account for roughly a third of the S&P 500’s total market value. The group includes Nvidia (NVDA), Alphabet (GOOGL), Amazon (AMZN), Apple (AAPL), Meta (META), Microsoft (MSFT), and Tesla (TSLA). When those names move, the index moves, and their daily swings can make the benchmark look healthy while the other 490-plus members sit flat or lower.
Burry described the pattern as “a very narrow top” and a “caldera.” His base case is not a single dramatic crash. He expects the market to drift lower, sometimes sharply, with the leaders pulling back while more beaten-down stocks recover ground. He says that rotation played out only once in the four decades of market tops he examined: in 2000.
Why the 2000 playbook matters now
The S&P 500 peaked in March 2000 and did not find a bottom until October 2002. The technology stocks that had driven the index to record highs lost more than half their value in the 18 months that followed. Stocks already in deep drawdown territory before March 2000 did not follow them lower. Several posted gains during that same stretch. Burry went back through every major market top since 1985. He found that rotation only in 2000.
Burry has named Cisco (CSCO) as the closest historical analog to Nvidia today. Cisco rose roughly 3,800% over the five years leading up to the March 2000 peak and briefly became the most valuable company in the world. Its stock then fell more than 80% and never recovered to that level. Burry describes Nvidia as filling the same role: the picks-and-shovels company at the center of a broad technology boom.
He is not alone in noticing the parallel. Billionaire investor Paul Tudor Jones has said the current environment feels like 1999, though he thinks the rally could extend another year or two. The comparison to 2000 is not a fringe view. The debate is over timing.
Burry has also noted that the top-performing Nasdaq-100 stocks ran further in the 12 months leading up to his analysis than the equivalent group did in the final stretch before the dot-com peak. That kind of concentrated, late-stage surge has historically marked the end of a move, not its middle.
Burry described the pattern as “a very narrow top” and a “caldera.” His base case is not a single dramatic crash.Astrid Stawiarz / Getty Images
Where money could go if leaders stall
Burry is not only watching. He is positioning around the thesis. In a late-September post on his Substack, Cassandra Unchained, he disclosed short positions in Micron (MU), Nebius, Palantir (PLTR), and a semiconductor exchange-traded fund. In the same note, he added long positions in five stocks sitting well below their 52-week highs: QXO, Build-A-Bear Workshop, Sprouts Farmers Market, Birkenstock, and MercadoLibre (MELI).
He included a caveat. Today’s beaten-down stocks are not as cheap as their 2000 counterparts were. The median valuation among those names is substantially higher than it was when Burry ran the same analysis on the 2000 period. He was direct about it: this is not shooting fish in a barrel.
The 10-year Treasury yield reached 5% in late September. That level had not been seen since 2007. Short-duration bonds and cash equivalents pay well at that rate. Longer-dated bonds are a different story. Their prices drop as yields climb, and the effect is larger the further out the maturity runs.
Michael Burry:
Michael Burry just made another bold move on Wall Street
Michael Burry doubles down on his stock market stand
For most investors, the practical question is simpler than any of those positions. If AI-related stocks have grown to a large share of a portfolio, including through broad index funds, that share is worth examining. Trimming toward other sectors or adding some bonds is a risk management step. It does not require a prediction about when or whether a 2000-style break happens.
Why selling everything would be a mistake
Burry’s direction has often been right. His timing has not. The S&P 500 gained considerably after the short position he disclosed in August 2023 and kept climbing well beyond that. Even Burry has said his trades are not appropriate for most investors to replicate.
Burry has not said this is 2000. He wrote on his Substack: “more like 2000 than any time since, but not 2000 yet.” His Fresh Drawdown Ratio, a narrower version of his main gauge that strips out stocks still dragged down by peaks from prior years, sits at about half the reading it posted in March 2000. The picture is moving in that direction. It is not there yet.
Nvidia, Alphabet, Amazon, Apple, Meta, Microsoft, and Tesla are not Cisco. Cisco in 2000 was selling networking hardware to internet companies burning borrowed capital. Most of those companies failed, and Cisco’s revenue collapsed with them. The Magnificent Seven sell products and services across a much wider range of paying customers. That is a meaningful difference when weighing the parallel.
Anyone holding a broad index fund already has concentrated exposure to AI through the Magnificent Seven’s weighting in the S&P 500. Those seven names make up roughly a third of the index. If they have appreciated faster than the rest of a portfolio, the actual weighting is higher than that. Trimming toward other sectors or adding bonds is a response to that concentration. Burry has made the case for the risk. His track record on timing is the open question.
Related: Michael Burry doubles down on his stock market bet