Your "Diversified" Index Fund Is 8% One Chipmaker

If you own an S&P 500 index fund, roughly eight cents of every dollar you put in goes to one company. Not to a sector, and not to a basket of chips — to Nvidia, and to no one else. Nvidia now accounts for about 8% of everything the index is worth, up from 6.5% within the past year and from 0.35% not long ago.

Here is what that number does to a beginner's mental model, because it should break it.

The index is not a market, it's a feedback loop

Start with the mechanic, because the whole surprise lives there. The S&P 500 is weighted by market capitalization — the bigger a company's market value, the more it owns of the index. If every stock were equal, Nvidia would hold 0.2% of the index, one slot out of 500. At 8%, the market has chosen to weight one company forty times heavier than a typical member.

Weighting by market cap does not just measure a company's size. It amplifies it. When Nvidia's price rises, its weight rises, which means a bigger share of every new index-fund dollar flows toward Nvidia, which pushes its price up again. The index is a machine that rewards its own winners with more buying. This is how a company that dragged along the bottom of the benchmark during the 2000s and 2010s became nearly a tenth of it.

And it is no small club. Nvidia plus the next nineteen biggest names now make up about half the entire index. The ten largest companies alone weigh in near 41% — a level of concentration the index has not seen since the mid-1960s. An investor who believed they bought "the whole economy" has, in practice, bought a bet whose direction is set by a handful of AI-exposed stocks. Passive funds reinforce it: more than $40 of every $100 going into S&P 500 funds lands in just those ten companies.

The hard part: the monster is real

Here is the natural objection, and it deserves a straight answer rather than a dismissal. This is not a hollow dot-com valuation. Nvidia's latest quarter was staggering: roughly $96 billion in revenue, up 106% from a year earlier, with $89 billion of that from data-center chips alone. It is the largest beneficiary of the biggest capital-spending cycle in corporate history. Its customers — Microsoft, Amazon, and their peers — are projected to spend more than $700 billion on AI infrastructure this cycle, and Nvidia's chief executive says "AI has reached its inflection point. Now, compute is revenue."

The company's own forecast is that it grows about 70% in the fiscal year ending January 2028, and it has already signed up partners like AWS to deploy two million additional GPUs in 2027 and 2028. So the eight cents is not resting on a story. It is priced off revenue that is actually landing.

But this is exactly where the concentration bites. The point was never that the earnings are fake. The point is that real earnings do not make an index diverse. Say Nvidia's thesis holds — the $5.5 trillion record it set in May, after the trillion-dollar summer pullback, proves nothing either way about what an AI capex slowdown would do to a benchmark that leans on it for 8% of its value.

What history actually says

The instinct is to conclude "concentration is dangerous, therefore the index is about to fall." The data does not support that leap. Study the last time the top ten approached half the index — the mid-1960s — and the cohort's fate was "underwhelming": several heavyweights eventually entered bankruptcy, and over the following sixty years the biggest names were not the ones that carried returns. The index still rose, because roughly 490 other companies kept producing winners that simply had not been big yet.

The lesson cuts against both panic and complacency. Concentration has only a mildly positive, historically unreliable link to what happens next. It neither forecasts a crash nor justifies ignoring the risk. What the mid-1960s story does show is that capital-weighting quietly solves its own problem — today's giants fade in weight while tomorrow's leaders, which start as relative minnows, grow into the benchmark — but only over decades, not over the next earnings call.

What you are actually buying

So where does the reader land? Not with a prediction about Nvidia's next quarter. With a corrected picture of the product they already hold.

An S&P 500 fund is no longer a diversified bet on the American economy. It is a leveraged bet on one mechanism — AI capital spending — with the margin driven by a single company whose weight is already 8% of the whole. That bet has produced enormous returns, and its underlying business is real. Defending it is not the same as understanding it.

The judgment is what happened, not what comes next. If you own index funds, you own this concentration whether you chose it or not — the choice to buy a broad market fund was also a choice to become an AI investor with no way to opt out. Nvidia earns its weight with genuine revenue, so this heavy lean is not a scam and not obviously a bubble. But diversification in the index is gone, replaced by a single bet that is only as safe as one company's ability to keep selling chips faster than the world can build data centers around them. Know that you bought it — because an index that silently owns 8% of one company no longer diversifies you from anything except the rest of the market.