The S&P 500 hovers near highs, yet 60% of its constituents have fallen into a technical bear market, with the Magnificent Seven tech giants single-handedly propping up the index. Hedge fund legend Paul Tudor Jones has issued a rare, all-out bearish call, warning that the market looks like a replay of 1999 and that 2027 could be the worst year for stocks since 2002. The AI boom, earnings concentration, and supply pressures harbor hidden dangers, while cloud giants' earnings will set the tone—when will the market inflection point arrive?
Author: Wall Street CN
The S&P 500 is a stone's throw from its all-time high, but beneath the surface, undercurrents are stirring: 60% of its constituents have already fallen into a technical bear market, leaving only the Magnificent Seven tech giants to prop it up on one leg. This inevitably brings to mind a scene from the past—1999.
Paul Tudor Jones, founder of Tudor Investment and a legendary hedge fund figure, has issued a rare, fully bearish signal on US equities: current market conditions resemble 1999, roughly a year before the dot-com bubble burst in 2000. He predicts 2027 will be "the worst" year for the stock market "since 2002," and he will short US equities for the entire year.
Before this, reports from Goldman Sachs, Morgan Stanley, and BTIG all pointed to the same year. Notably, both the bulls and the bears are citing 2000.
The index is at highs, yet 60% of stocks are already in a bear market
Michael Cembalest of JPMorgan Asset Management laid out a set of figures: 60% of S&P 500 constituents have fallen more than 20% from their respective all-time highs, while the index is still hovering near its high—extremely rare since 1990. Over the past three years, only 30% of constituents have outperformed the index, the lowest since 1991. The last time this reading appeared was in 1999 to 2000. Goldman Sachs was even more direct: strategist Tony Pasquariello's report was intended to defend the bulls, yet he wrote this line: "The median S&P 500 stock is currently down 15% from its 52-week high, the worst reading since the dot-com bubble."
The last leg is the Magnificent Seven
The index is still rising because a small handful of stocks are rising.
Another set of figures from Michael Cembalest: since January 2024, 42 AI-related stocks have contributed about 67% of the S&P 500's price return, 58% of its earnings growth, and 81% of its capex and R&D spending growth. Around forty stocks are carrying two-thirds of the index. Jonathan Krinsky of BTIG titled his report "The Last Gatekeeper." His judgment is blunt: the Magnificent Seven are the only sector in the S&P 500 that has yet to show signs of fatigue, and also the most crowded trade; once it turns, it will likely be the fuse for a sustained index decline. The numbers are on his side. The Roundhill Magnificent Seven ETF (MAGS) has already hit an all-time high, while surrounding sectors are broadly pulling back.
On Tuesday, against the backdrop of semiconductor stocks falling again, the Magnificent Seven rose about 1% collectively, single-handedly propping up the S&P 500 index. Hedge funds are using these seven as a safe haven, hiding from geopolitical risk and Treasury yields above 5%. Too many people are hiding there, and concentration has hit a record—the Magnificent Seven account for about 22% of Goldman Sachs' US equity prime brokerage exposure.
Goldman Sachs' Rich Privorotsky summed it up in one line: "The market has become extremely overbought in one corner while effectively being short everything else." Is a safe-haven asset that everyone holds still a safe-haven asset? The Philadelphia Semiconductor Index has closed lower for five consecutive trading days, its longest losing streak in nearly 17 months.
Sounds grim, but as of Monday's close, the SOX is still up about 72.8% year-to-date, with Micron up over 240% for the year. Down five days, still at highs—this is probably what 1999 looked like.
Jonathan Krinsky also overlaid the Philadelphia Semiconductor Index (SOX) trajectory from 1999 to 2000 for comparison, and the degree of overlap is "unsettling." The SOX has now pulled back nearly 18% from its June high, which alone is nothing, but against the 1999-2000 curve it looks ugly. By his analogy, semiconductors will first have a brief rebound (CPI is the catalyst), then accelerate downward in late October. What is late October? The window when major cloud computing giants report earnings in concentrated fashion.
Earnings are also peaking, but no one wants to admit it
Goldman Sachs itself calculated that S&P 500 earnings growth of 27% to 28% in the third quarter is the peak of this cycle, slowing quarter by quarter thereafter.
Pasquariello said 2027 "will be a complex year." Profits are still concentrating in one direction. Since March 2023, the trailing earnings of 55 "AI supply chain companies" have grown about fourfold, while 190 "AI high-adoption" companies have grown only about 25%.
Those selling shovels are making a fortune; those mining haven't struck gold yet. This script was played out in 1999.
The tech sector's contribution to US GDP growth rose from nearly zero in 2022 to about one-third, but pre-built data center planned capacity has fallen from a peak of about 62 gigawatts to about 49 gigawatts. The only expanding capex engine is starting to shrink its plans.
The real killer move is in 2027
The 1999 story has a second half, and that is the force that truly pushed the market down—supply. The bull market of the past four years has a rarely mentioned pillar: corporate buybacks and M&A continuously compressing the float. This logic is now about to reverse.
Paul Tudor Jones said it plainly on CNBC: the current US equity supply-demand setup is "the worst since 2002," like 1999, the year before the bubble burst. Large IPOs from Anthropic and OpenAI are coming, and SpaceX's offering has just been completed. In addition, many companies will launch secondary offerings to raise funds for AI buildout. Jones believes these factors combined will create enormous equity supply pressure in 2027.
He announced he will short US equities for the entire next year—"there's really no value there." A Goldman Sachs internal report on 2027 US equity supply and demand gives this wave of supply a magnitude: $600 billion in primary market issuance, of which $175 billion is IPO; buybacks plus M&A provide about $2 trillion in buying; but lock-up expirations could release $1.7 trillion in supply.
$600 billion plus $1.7 trillion is $2.3 trillion, against only $2 trillion on the other side. The incremental net supply of US equities could be the largest since 2000.
If most insiders choose to cash out after lock-up expiration, the buyback force that has cushioned every decline over the past decade may, for the first time, be overwhelmed by supply. The last time net supply of this magnitude appeared, Pets.com was still in business.
Money itself is also getting more expensive. Tencent Holdings issued dollar bonds and dim sum bonds in Hong Kong on Tuesday, with the 10-year dollar bond's initial price guidance about 1.15 percentage points above Treasuries; SoftBank issued $4.5 billion in high-yield bonds last month, paying up to 9.75% interest; Firmus, backed by Nvidia, shelved its planned $5 billion IPO in Australia last week.
BlackRock's research team simply downgraded its US equity outlook to "neutral" on Monday, turning bullish on short-term Treasuries instead; Broadcom CEO Hock Tan said capital could become one of the biggest constraints on AI industry growth.
The first reaction is in the bond market
The 10-year Treasury yield has already broken through 5%. If yields don't come down, the rationale for defensive money staying in seven big stocks holds; once they retreat, one reason disappears. Goldman Sachs' short-term strategy is to short fixed income.
On the US equity side, the short-term judgment is less pessimistic. Goldman Sachs believes the "tactical positioning environment is net positive," with technicals still having room to improve before the midterm elections. Jones is looking at 2027, a different time scale. Another market signal is the rotation between semiconductors and software.
The relative strength ratio of software versus semiconductors (IGV/SMH) has reclaimed its 200-day moving average for the first time in 16 months. If it continues to break above the summer high near 0.20, leadership will shift from semiconductors to software, and software is precisely not where the market is heavily positioned. Crude oil is a rare clear long in this cycle.
Jones holds a long position, reasoning that the US-Iran conflict will be hard to resolve in the short term: "I think it can rise another $20 to $25." WTI is currently around $94 per barrel; another $20 to $25 would break above the roughly $113 closing high in 2026.
He warned that if oil really moves like that, the stock market will "absolutely, thoroughly hate it." When it will blow up has two thresholds. The trigger condition is late-October cloud giant earnings: once capex guidance slows, or lock-up expirations exceed expectations, the Magnificent Seven will most likely catch down.
The invalidation condition is third-quarter earnings beating expectations, fourth-quarter guidance not being revised down, and actual net supply significantly below estimates—then this "supply crushes buybacks" thesis must be overturned.
Watch cloud giant earnings nextOver the next two weeks, the key thing to watch is the earnings season for cloud computing giants, especially their capital expenditure guidance, which is the source of the entire AI profit chain. There is also something easily overlooked: bond issuance rates. The coupons paid by Tencent and SoftBank are a thermometer for AI financing costs. Not all signals point in the same direction, either. After Meta released a new model on Monday, its stock rose about 3.4% against the market trend. What the market rejected was indiscriminate bets on infrastructure, not AI itself. At the end of Pasquariello's report, a different picture is painted:
Since the low in March 2009, the total return of the S&P 500 has risen 16-fold. During this journey, there has been only one recession, and it lasted only about two months.
Even so, the reading from the University of Michigan consumer sentiment survey today is lower than in March 2009—the just-released figure (46.3) is the second-lowest reading since records began in 1952.
Pasquariello calls this divergence "shocking and somewhat puzzling." During the same period, those who held stocks got 16 times their money. What others got was simultaneously rising home prices, rents, and food prices, along with the reality that after the 10-year U.S. Treasury yield broke above 5%, every mortgage became a luxury. ...
What 1999 left to later generations was not the sentence "the bubble will definitely burst," but that, standing one year before the bubble burst, the vast majority of people were still bullish.