Thursday looked like a highlight reel from 2023. Nvidia closed up roughly 8% after fiscal Q2 revenue of $96.2 billion, the Nasdaq Composite gained about 1.5%, the S&P 500 added about 0.7% — and the Dow managed 0.2%. One stock, one sector, most of the move.
The year has not worked that way. Through Thursday the S&P 500 is up 12.13%. The equal-weight version of the same 500 companies is up 15.76%. That is a 3.6-point gap in favour of the boring one, and it is the first sustained stretch in years where owning every constituent in the same size has beaten owning them in proportion to their market value.
Money noticed. The Invesco S&P 500 Equal Weight ETF crossed $100 billion in assets for the first time this month, after pulling in more than $12 billion so far this year.
Our take: The interesting part is not that equal weight is winning. It is what has to be true for it to win. The cap-weighted index carries roughly 40% of its weight in ten names; Nvidia alone is about 8% of the S&P 500 and 14% of the Nasdaq-100 at a $4.6 trillion market value. For the other 490 to out-run that arithmetic, the mega-caps have to go sideways while everything else grinds higher — which is exactly what happened. The Magnificent 7 were roughly flat through the first half of 2026 while the index gained 9.3%. Equal weight is not beating the mega-caps. It is beating a mega-cap group that stopped compounding and started spending.
Flat is not the same as broken
The Mag 7 did not stumble on demand. They stumbled on capital expenditure. Every dollar committed to data centres, power and silicon lands immediately in the cash-flow statement and only later, if ever, in the revenue line. Nvidia's own quarter is the cleanest illustration of the other side of that trade: data-centre revenue of $89.02 billion, up 116.6% year over year, is somebody else's capex budget arriving as somebody's sales.
That is the split the equal-weight number is really measuring. The companies writing the cheques have been marked down for writing them. The companies cashing them have been marked up. And the 490 names doing neither — industrials, insurers, regional banks, healthcare — got re-rated by default when money that used to sit in seven stocks went looking for somewhere else to be. We flagged that chips-to-banks rotation earlier this month; the equal-weight line is the same story with an asset number attached.
The gap under the gap
There is a structural argument underneath the performance one, and it is older than this year. Through 2025, the top ten names accounted for roughly 41% of the index's weight but were expected to generate about 32% of its earnings. That spread has widened meaningfully since 2015. You can read it as evidence the biggest companies deserve a premium multiple, or as evidence that index buyers are paying nine points of weight for earnings that are not there. Both readings have been profitable at different times.
What is not arguable is the concentration itself. If you hold an S&P 500 fund, a total-market fund, a Nasdaq fund and a growth fund, you do not hold four things. You hold roughly the same ten companies four times over, with a slightly different tilt each time. Most people have never checked the number — our concentration audit playbook walks through how to calculate your real single-name exposure in about half an hour.
What to watch
- Friday morning. Fed chair Kevin Warsh delivers his first Jackson Hole keynote at 10:00am ET. Breadth trades are more rate-sensitive than mega-cap tech, because the 490 carry more debt and less cash. A hawkish surprise hits equal weight harder.
- Whether the gap survives a down tape. Equal weight has led during a rising market with flat mega-caps. It has not yet been tested by a broad drawdown, where the diversification argument and the quality argument point in opposite directions.
- Nvidia's index weight. At about 8% of the S&P 500, a 10% move in one stock is roughly 80 basis points of the index before anything else trades. That number is the whole debate in one line.
- Flows, not just returns. The $12 billion that arrived this year showed up after the outperformance, not before it. If it leaves the same way, crossing $100 billion will read as a top rather than a turn.
None of which tells you what to own. It tells you what you already own — the harder question, and the one almost nobody asks until a single name is 8% of an index and rather more than that of a portfolio nobody meant to build.