60% of stocks have fallen into a bear market, the seven tech giants are holding up the S&P alone! BTIG: US stocks may be just a "gentle push" away from collapsing

The S&P 500 Index is just one step away from its historical high, but beneath this calm surface, the market is already experiencing severe internal divergence. Currently, the only pillar supporting the index is the "Magnificent Seven" tech giants—which is precisely where the greatest risk lies for the market.

In the latest report "Last Man Standing," technical analysis firm BTIG’s Jonathan Krinsky warns that the Magnificent Seven tech giants are the only sector within the S&P 500 that has not shown significant weakness and are also the most overcrowded holdings in the market right now. He points out that once this trend starts to reverse, it is likely to become the trigger for a sustained decline in the S&P 500.

Meanwhile, data from Michael Cembalest at J.P. Morgan Asset Management shows that more than 60% of S&P 500 constituents have fallen more than 20% from their respective all-time highs, entering a technical bear market, even as the overall index still hovers near record highs.

This stark divergence highlights the market’s growing fragility. Analysts note that when defensive trades and growth trades are concentrated in the same seven stocks, the market has virtually lost the room for rotation; and the ongoing decline of the semiconductor sector is following the pattern of the burst of the 2000 tech bubble, suggesting real pressure might be unleashed later this month.

The Magnificent Seven Alone Can’t Hold Up the Market; Crowded Trades Hide Dangers

Krinsky’s key judgment is straightforward: the Magnificent Seven are currently the only major sector still hitting new highs, but they are also the most concentrated holdings.

Market data show that the Roundhill Magnificent Seven ETF (MAGS) has already surpassed its 2025 and May highs, reaching a new record, while surrounding sectors have mostly pulled back. On Tuesday, as semiconductor stocks declined again, the Magnificent Seven as a whole rose about 1%, single-handedly propping up the S&P 500 Index.

However, this sense of "safety" is precisely the problem. Hedge funds see the Magnificent Seven as a safe haven to escape 5%+ Treasury yields and geopolitical risks, flooding into these stocks and pushing concentration levels to historical extremes. Krinsky bluntly points out: When defensive trades and momentum trades focus on the same seven stocks, the market has nowhere left for rotation. An ‘insurance asset’ that everyone owns is no longer an insurance asset at heart, but rather an attractively-packaged crowded trade.

Semiconductor Performance Mirrors 2000, Accelerated Decline Could Come at Month's End

Another warning signal for BTIG comes from the semiconductor sector. Krinsky has been comparing the Philadelphia Semiconductor Index (SOX) to its trend from 1999-2000 since this summer, calling the resemblance "unsettling."

According to this analogy, the semiconductor sector is now near a short-term bottom, about to see a brief rebound—where the CPI data release could be a catalyst—followed by accelerated decline in late October. This timing closely coincides with the earnings release windows of the major hyperscalers.

Krinsky also notes another signal: the relative strength ratio between software and semiconductors (IGV/SMH) has, for the first time in 16 months, recaptured its 200-day moving average. He believes that if this ratio further breaks above the 0.20 mark near summer highs, it will "confirm a major turning point," meaning market leadership will shift from semiconductors to software—which is not the current heavyweight position of most portfolios.

42 AI Stocks Account for 60% of Gains, Others Are Just Supporting Cast

In his latest report "Califoreboding," Michael Cembalest of J.P. Morgan Asset Management, under the heading "One Trick Pony," uses data to reveal the extreme degree of market concentration.

Since January 2024, 42 AI-related stocks have contributed about 60% to 80% of the S&P 500’s earnings growth, price returns, and capex growth. Specifically, these 42 stocks accounted for 67% of the index’s price return, 58% of earnings growth, and as much as 81% of the increases in capital expenditures and R&D spending.

Meanwhile, even though the S&P 500 is near its all-time high, more than 60% of its constituents have fallen by more than 20% from their respective peaks. According to Cembalest’s statistics, since 1990, it is extremely rare for such a large proportion of individual stocks to have pulled back so far when the index itself is less than 1% from its all-time high; the reading on September 25 sits at the extreme left side of the historical distribution. Furthermore, only 30% of S&P 500 constituents have outperformed the index in the past three years—the lowest share since 1991, rivaling only the trough from 1999 to 2000.

AI Profits Flow to ‘Shovel Sellers,’ Users Have Yet to Find Gold

Cembalest’s data further reveals another side of the AI boom: profits are highly concentrated among the suppliers of AI infrastructure, not the users.

Since March 2023, the rolling earnings of 55 "AI supply chain" companies have grown by about fourfold; meanwhile, the earnings of 190 "AI highly-adoptive" companies have grown by about 25%; and earnings for those with the lowest AI adoption have stayed flat or even declined slightly.

This pattern is very similar to the dot-com bubble: those selling shovels are making outsized profits, while the miners have yet to find their gold. The concentration effect has also spread to the real economy—the tech sector's contribution to US GDP growth has risen from almost zero in 2022 to about one-third, yet only about 20% of US states are growing faster than the national average—close to the lowest since 1979.

Regarding fixed investment, data center and power infrastructure spending has increased by 38% since January 2024, while manufacturing construction has dropped by 28%, commercial real estate by 13%, and residential construction by 6%. More importantly, pre-construction capacity planning for data centers has dropped from a peak of about 62 GW to about 49 GW, indicating that the only engine of capital expenditure growth is already starting to contract in terms of future planning.

Bulls Still Have Support, But Profit Foundations Are Narrowing

Cembalest is not outright bearish. He points out that since January 2024, the rise in global tech stocks has primarily been driven by earnings growth, not valuation expansion, and that "in my experience, a market driven by earnings growth rarely sees sharp declines." Additionally, despite rising short-term interest rates, corporate net interest expenses remain near historical lows as a share of profits.

However, these profits are highly concentrated—mainly in AI supply chain companies, whose gains are largely tied to capex spending by hyperscalers, which increasingly relies on debt financing. The profits themselves are real, but the concern is whether these customers can sustain their spending. According to Goldman Sachs estimates, Q3 may see the dual peaks of both S&P 500 earnings growth and hyperscaler capex growth.

Synthesizing BTIG and J.P. Morgan’s analysis, the current market picture is both clear and tense: 42 stocks have contributed two-thirds of the overall return and four-fifths of the growth in capital expenditures; 60% of the constituents are already in bear markets; only 30% of stocks are outperforming the benchmark; defensive and growth trades are concentrated in the same seven names, with hedge fund positioning hitting historical highs; the semiconductor sector has followed the same decline path as in 2000, and the acceleration phase is likely to come around late this month, coinciding with hyperscaler earnings season.

Krinsky’s conclusion is restrained: "Fighting new highs is never easy." The charts for the Magnificent Seven still point upward. But as he warns: when the market is left standing on just one leg, it doesn’t take much force to knock it over.