The S&P 500 rose 131% in four years: structural rift under AI and the end of a bull market

According to Woofun AI, the S&P 500 index recorded an impressive 131% increase over four years, but under this bright macro image, the ultimate structural divergence is hidden: up to 84% of the constituent stocks have pulled back more than 10% from their respective highs, while only 42 AI-related stocks contributed 67% to the index's increase, and the market capitalization share of the three tech giants Apple (AAPL.US), Microsoft (MSFT.US), and NVDA.US (NVDA.US) has soared to 22%.

This contradictory picture of a new high index and an extreme narrowing of the market width indicates that this round of the US stock bull market is standing at a tense crossroads, and its roots are being supported by fewer and fewer companies. Since bottoming out on October 12, 2022, the total cumulative return of the S&P 500 has reached 131%, equivalent to an annualized yield of about 23.3%, which is about three times the historical market average. During this period, Nvidia's stock price soared by about 1875%, and Meta (META.US) rose by about 488%, becoming the core engine driving the index forward.

However, the current market breadth of the S&P 500 index — that is, the share of rising stocks among the index's constituents — has fallen to a historic low. Historical experience shows that whenever there is such a serious divergence between market breadth and index trends, the S&P 500 often hits the top within the next few months.

Meanwhile, continued pressure on the bond market has cast a shadow over the risk asset outlook, and US Treasury yields have risen to 24-year highs. The core contradictions in the market are already clear: profit growth and investment boom in the field of artificial intelligence still form the main support, but the high interest rate environment, energy inflation risk, and extremely concentrated market structure have left the market with little room for fault tolerance, and any failure in expectations may cause sharp fluctuations.

The biggest driving force of this bull market has always been fanatical bets on artificial intelligence, but the wealth effect is being concentrated on a few leading companies at an accelerated pace, forming a winner-take-all pattern. According to market research firm The Kobeissi Letter, 42 AI-related stocks have contributed about 67% of the S&P 500 price return since the beginning of 2024. These 42 stocks contributed about 22 percentage points to the index, while the remaining 458 constituent stocks combined contributed only 11 percentage points. This huge gap in contributions revealed serious imbalances within the market. The concentration problem is worsening. The three companies Apple, Microsoft, and Nvidia now together account for about 22% of the total market value of the S&P 500, which is a sharp increase from about 14% at the end of 2022.

Meanwhile, currently about 84% of the constituent stocks in the S&P 500 have fallen by more than 10% from their respective historical highs, and are already in a correction range in the technical sense. This data appears at a time when the index itself is still at an all-time high, which is particularly dazzling. Nationwide's chief market strategist Mark Hackett summed up this situation as “the core tension in the market”: profits and AI investment are still strong supports for growth stocks, but high long-term interest rates, energy-driven inflation risks, and extremely narrow market breadth have left little room for fault tolerance in the market.

Data compiled by Woofun AI shows that this concentration risk is becoming the sword of Damocles hanging over the top of the market. Once the performance of leading companies falls short of expectations, indices lacking support from other sectors will face tremendous downward pressure.

As the bull market enters its fifth year, some investors are beginning to worry about whether it is nearing an end due to the “old dragon clock,” but historical data does not support this assertion. Of the 38 rounds of the bull market since 1900, 33 have been shorter in duration than the current round. The average duration of all bull markets was 25 months, and the median was 22 months.

This data may seem alarming at first glance, but MarketWatch columnist Mark Hulbert cites research as pointing out that using “age” to predict the end of a bull market is a statistical fallacy—just like thinking that day causes night because night follows day. The real end to a bull market is often an economic recession (13 of the 22 bears since 1950), excessive investor excitement (such as the internet bubble), or external shocks (such as wars and pandemics). CNBC host Jim Cramer confirmed this from an investor's perspective: “We simply can't predict which will be a real sharp drop, nor can we predict when the rebound will take place. We've been able to escape countless times in four years, but every time we've lost our lives.

“Furthermore, historical data also provided a relatively optimistic answer to the question of whether the increase was already too high. Edward McQuarrie, professor emeritus at Santa Clara University, and William Bernstein, founder of EfficientFrontier.com, conducted statistics on the total monthly rolling four-year return of the US stock market since the end of the 18th century, and found that the current four-year cumulative increase of 131% did not constitute a historical anomaly. The two wrote, “This happens often... it's nothing at all, far from the extreme in history.” Even if we extend our perspective to the low point of the global financial crisis in March 2009, with 17.5 years as the statistical cycle, the performance of the US stock market has had many precedents in history where it has surpassed the current level by a large margin.

This means that judging that the bull market is coming to an end based on the increase itself lacks a strong historical basis.

The real risk in the current market is probably not the age and increase of the bull market, but rather whether interest rate pressure can be relieved and whether the market breadth can be repaired. Morgan Stanley (MS.US) chief investment officer Mike Wilson previously pointed out that the deterioration in market breadth is closely related to the hawkish tone of the Federal Reserve. Following the strengthening of the Federal Reserve's hawkish stance, interest rate expectations moved upward, and fluctuations in the bond market boosted US bond yields to a 24-year high, further suppressing the overall outlook for equity assets. He wrote, “If bond volatility calms down faster, market breadth is likely to catch up with index prices, driving both higher in the near future.

“The HSBC (HSBC.US) multi-asset strategy team shares a similar view. According to the HSBC report, if the US economic data shows signs that it continues to fall short of expectations, it is expected to drive the entire US bond yield curve downward, thereby supporting the valuation of various types of assets and contributing to the spread and rebalance of the driving forces within the stock market. HSBC strategist Max Kettner's team believes that after entering 2027, if the “blonde girl” combination is achieved, that is, growth is steady, inflation falls, and interest rates fall, profits from technology stocks and US stocks may return to strength at that time. However, the rise at that time may still be a round of highly concentrated structural markets, rather than a general rise across the board.

This outlook shows that market participants are closely watching macroeconomic policy shifts and changes in economic data to determine whether the bull market can spread from a narrow AI circuit to a wider sector.

Another core debate in the market is whether AI investment has entered the bubble phase, but fundamental data provides short-term support. The HSBC strategy team clearly refutes this. The report points out that the problem with current AI skepticism is that it is too long-term — using highly uncertain revenue predictions after two to three years to disprove the current price trend, the logic doesn't work. The more important recent sign is that free cash flow (FCF) downgrades for hyperscalers (hyperscalers) have begun to slow, and expectations for FCF improvements in 2027 are being formed. Semiconductor and memory companies have recently taken the lead again, and computing power prices have stopped falling and rebounded, showing the characteristics of the “Jevons Paradox” (Jevons Paradox) — that is, increased efficiency instead drives larger demand expansion, posing a potential positive catalyst for the entire technology sector.

Profitability also provided strong support. Currently, a record number of S&P 500 companies have issued positive profit guidance for the third quarter. Samsung Electronics previously predicted that operating profit for the third quarter is expected to reach 80 billion US dollars. If realized, it will set the highest single-quarter profit record in the history of a technology company. To a certain extent, this explains why in an environment of war, high interest rates, and economic uncertainty coexist, the index is still at an all-time high — capital is flowing to places where real money is being made. As the game between AI profit reality and future expectations intensifies, the market will find a new balance in structural rift, and this process will determine the final direction of this bull market.