Nvidia hits $3trn, and index funds are along for the ride

Nvidia's market value passed $3 trillion on 5 June, overtaking Apple to make it the second most valuable company in the world. As the shares rose, index-tracking exchange-traded funds and passive mutual funds had to buy heavily to keep up with the chipmaker's growing weight in their benchmarks.

Two weeks of fuel

The latest leg came from results. Late last month Nvidia reported another quarter of data centre sales far above a year earlier, guided higher again and announced a ten-for-one stock split that takes effect after trading today. Microsoft, the only company now worth more, is within reach. Three companies, all tied closely to the same bet on artificial intelligence, now make up a larger slice of the S&P 500 than whole sectors of the American economy.

The numbers behind the rally are real. Revenue for the quarter to April came to about $26 billion, more than triple the figure a year earlier, and the data centre business accounted for the great majority of it. The customers are concentrated too. Microsoft, Alphabet, Amazon and Meta have all raised their capital spending plans to build artificial intelligence capacity, and a large share of that money ends up as Nvidia's sales. The biggest companies in the index are therefore, to a considerable extent, each other's revenue.

Index funds chase the winner

When one stock grows this fast, indices weighted by market value create a feedback loop. Every dollar that goes into a passive fund sends an outsized share into the biggest, fastest-rising companies, pushing their valuations up whatever the fundamentals say. Passive investing, sold as the neutral option, now amplifies the market's concentration in a few mega-caps.

The scale of the flow has grown with the industry. By Morningstar's count, passive funds now hold more American fund assets than active ones. Pension contributions and retirement savings flow into them every payday, automatically and without regard to price. Sector funds add a second channel: those that cap the weight of any single holding must trade heavily at their quarterly rebalancing when the ranking among their largest stocks changes.

Why small flows move big prices

The textbook view holds that prices should barely respond to buying that carries no information, because other investors will sell to anyone paying too much. Recent research suggests the market is less elastic than that. Xavier Gabaix and Ralph Koijen have estimated that a dollar moved into equities raises aggregate market value by about $5, because few investors are both willing and able to lean against sustained flows. If that holds for individual stocks as well, persistent passive buying of the largest names has a price effect well beyond its size.

Precedents from index rules

The market has seen mechanical buying move a single stock before. When Tesla joined the S&P 500 in December 2020, index funds had to buy tens of billions of dollars of its shares, and the stock rallied hard between the announcement and the inclusion date. In July 2023 Nasdaq carried out a special rebalancing of the Nasdaq-100 because a handful of giants, Nvidia among them, had grown too large for the index's own diversification rules. Both episodes show index providers and their trackers reacting to concentration after it has built up, rather than preventing it.

The split adds a quieter mechanism of its own. The Dow Jones Industrial Average weights its members by share price rather than market value, which made a stock trading above $1,000 impractical to include. At a tenth of the price, Nvidia becomes a plausible candidate, and funds tracking the Dow would then have another reason to buy. Splits do not change a company's value, but they widen the pool of indices and retail buyers able to hold the shares easily.

The dot-com echo

The obvious historical comparison is Cisco, which briefly became the most valuable company in the world in March 2000 as the supplier of the networking equipment that the internet build-out needed. Its sales kept growing for a while after that, but the share price lost most of its value over the next two years as customers cut orders and investors repriced the whole sector. The lesson was that selling the picks and shovels of a boom is a fine business until the buyers decide they have bought enough.

The differences matter as well. Nvidia is far more profitable than Cisco was, and its valuation relative to expected earnings is lower than the telecoms-equipment makers reached at the peak. Its customers are some of the richest companies in the world rather than debt-laden start-ups. The parallel is about the cyclicality of capital spending, not about fraud or froth, and it suggests the risk lies in the order book rather than the balance sheet.

The case for the defence

The strongest objection is that a fund weighted by market value does not need to trade when a holding rises; its weight grows by itself. On this view passive funds are price-takers, and active managers and hedge funds set the price of Nvidia. That is correct for the existing stock of assets. It says less about the stream of new money, and passive funds get the lion's share of it, while active managers who lag their benchmark lose clients and are pushed to own more of the winners too.

The losers from concentration are equal-weighted indices and active managers, both of which have lagged the headline index badly since the start of last year. The winners are anyone who owned the giants early. The risk falls on savers who believe a broad index fund spreads their money widely, when a large share of it is now riding on one theme.

The test is what happens when news disappoints. Nvidia's next results come in late August, and the hyperscalers report on their capital spending a month earlier. Fund-flow data and any slowdown in retirement contributions would show whether the mechanical bid is fading.

Index investors own a large and growing bet on Nvidia, whether they meant to or not. If the momentum turns, the same mechanical flows that pushed the shares up could help drag them down. The first real test comes with the hyperscalers' spending plans in late July, and a single disappointing capital-spending forecast from any of the four big buyers would hit Nvidia, the index and every passive saver at the same time.