10 stocks account for 70% of the S&P 500’s gains amid structural market imbalance.

Author: Shenchao TechFlow

The apparent prosperity of the S&P 500 index is masking the most extreme structural divide in U.S. stocks in nearly a quarter of a century.

According to the latest research report by Charlie McElligott, Head of Cross-Asset Strategy at Nomura Securities, and data compiled by The Kobeissi Letter, since the S&P 500 index rose 23% on March 30, just 10 mega-cap stocks accounted for 70% of the entire market’s gain, while the remaining 490 components contributed only 30%.

On this highly distorted contribution chart, semiconductor and AI core assets dominate:

• Nvidia: Accounts for 13% of the index's return contribution (driving the market up by 3 percentage points), firmly holding the top position as the primary engine;
• Micron, Apple, Microsoft: each contributing 9% (each pulling the market up by 2 percentage points);
• AMD: contributes 8%;
• Alphabet (Google): contributes 7%;
• Intel and Amazon: each contribute 5%;
• Broadcom and Meta: each contribute 3%;
• The remaining 490 stocks combined contributed only 30% (collectively driving the market up by 7 percentage points).

The original news brief only revealed the extreme concentration of gains, but the Nomura report further exposed a more severe underlying micro-ecosystem—“the index has stagnated, while beneath the surface, a rolling bear market has already taken hold.”

1. Breadth collapse: One-sixth of the components have halved in value.

Although the S&P 500 is just one step away from its all-time high, the extent of damage at the component stock level is alarming:

• 85% of the component stocks have retraced more than 10% from their respective all-time highs (entering a technical correction);
• 59% of the component stocks have declined by more than 20% (entering bear market territory);
• 41% of the component stocks have declined more than 30%;
• 26% of the component stocks have declined by more than 40%;
• Even 17% (nearly one-sixth) of the component stocks fell more than 50%.

Over the past one and a half months, interest-rate-sensitive sectors have experienced broad sell-offs: healthcare (-3.5%), consumer discretionary (-3.6%), utilities (-4.6%), industrials (-5.8%), financials (-7.7%), materials (-7.9%), and real estate REITs (-9.8%). Across the entire S&P 500, only TMT (technology, media, and telecommunications) and energy have barely posted positive returns in the past month. The so-called "bull market" is being driven solely by a handful of AI favorites.

1. 25-year low correlation: Beware of a repeat of the "volatility apocalypse"

Nomura's derivatives data shows that the 1-month realized correlation of the S&P 500 has fallen to the 0.7th percentile over the past year, while implied correlation is at the 1.8th percentile.

Analysts from Goldman Sachs and Nomura simultaneously noted that in the past 25 years, extreme scenarios in which correlation dropped to such low levels have occurred only twice: once just before the global financial crisis in 2007, and again in early 2018 during the “Volmageddon.”

The apparent low volatility of the broader market (S&P 10-day realized volatility at just 7, with a monthly range of only 0.8%) is being suppressed by a "market maker positive gamma trap" driven by massive selling of same-day expiring (0DTE) options and quantitative implied volatility premium strategies (QIS); however, the average volatility of individual components stands at 8.9%, creating a gap of 8.1 percentage points—the highest in the 95th percentile over the past 30 years. Risk has not been eliminated; it has merely been confined to the individual stock level.

This artificial prosperity, sustained by only ten stocks, rests on a fragile assumption: that high capital expenditures (Capex) in upstream computing infrastructure will rapidly generate blockbuster commercial returns downstream, which then spill over and boost the productivity of the remaining 490 traditional companies.

However, the key premise of this chain is beginning to show cracks:

1. Rumor of OpenAI's revenue falling short of expectations: According to the Financial Times, OpenAI's actual annualized revenue is approximately $20 billion, a shortfall of over $500 billion compared to the nearly $70 billion previously projected to investors. This directly undermines the valuation rationale for hardware suppliers such as NVIDIA, Broadcom, and Micron.
2. Chain reaction of European energy and interest rate volatility: Structural shortages in European diesel refining capacity (Crack Spread) are highly correlated with one-year EUR interest rate volatility. Persistent macro interest rates maintained at a “higher for longer” stance continue to elevate borrowing costs for traditional non-AI industries, keeping valuations under pressure.

When capital has bet all its chips on the top 10 stocks, such as NVIDIA, the market's call skew has been pushed to a historical extreme of 99.6%—almost no one is buying downside hedging protection, and all institutions fear missing out. This one-sided, excessively crowded move, lacking broad-based support, often means that even the slightest crack in the earnings or core narrative of the leading stocks will dramatically amplify fragility at the broader market index level.