10 Stocks Drove 72% of the S&P 500's Gain This Year — The Concentration Risk Hiding in Plain Sight

TL;DR Around 72% of the S&P 500's rally this year came from the top 10 names, and those names are dominated by AI semiconductors. The sector trades ~60% above its 200-day moving average — a level BofA only matches to the Mississippi Bubble of the 1700s and the late-stage dotcom run. The index's downside risk is concentrated in a single theme.

The S&P 500 has 500 stocks. About 72% of its year-to-date gain came from the top 10. Those top 10 are dominated by AI semiconductors and data-center infrastructure names. The other 490 stocks did very little. That, more than any single macro headline, is the structural weakness I'd point to in this market.

~60% above the 200-day — where that sits historically

The semiconductor complex trades roughly 60% above its 200-day moving average. The 200-day is a rough proxy for the trailing one-year average price; being 60% above it is the rubber-band stretched close to snapping.

Bank of America compared the current setup to just two prior episodes: the Mississippi Bubble in 1700s France, and the late-1990s dotcom run. Neither ended quietly. Distance from the 200-day isn't a precise timing tool, but extreme readings are a real-world warning that risk has become asymmetric.

The market-breadth problem

Breadth asks how many stocks are participating. Healthy markets have wide breadth; brittle ones have narrow breadth. Right now, breadth is unusually narrow, and the narrowness is themed: AI accelerators, HBM memory, hyperscaler capex, the power and cooling infrastructure that sits underneath. If the theme wobbles, the index wobbles.

Two scenarios for the theme breaking

Scenario A: a single quarter of weaker AI capex guidance from a hyperscaler or a major customer cracks consensus. Multiples compress across GPUs, HBM, and the broader data-center supply chain.

Scenario B: macro breaks first — rates or inflation force long-duration assets to re-rate. The mechanism differs, but the outcome rhymes.

Either way, when the theme carrying the index stops carrying it, the index move can be fast. The Japanese Nikkei in 1989 started with the same fingerprint — a small set of leaders, an enormous distance from trend — and lost 80% on the way to a 30-year recovery.

I'm not telling you to dump semis

My actual takeaway is narrower: index exposure and concentrated single-name exposure are different risks. If you hold semis only through the index, the question is whether your overall index weight makes sense. If you hold them directly, the question is whether each name clears the filters in my crash-winner framework.

What I'm watching

Three confirms I'd want to see breaking in the same direction: the semiconductor ETF's deviation from its 200-day, the equal-weight S&P 500 (RSP) versus the cap-weighted S&P (SPY), and the 30-year Treasury yield. When two of three line up, the signal stops being noise. For the broader macro setup behind this, see my notes on the BofA Door to Doom report.