Nvidia becomes the world's most valuable company

Nvidia overtook Microsoft on 18 June to become the world's most valuable listed company, with a market value of $3.34 trillion. Two years ago that would have seemed absurd. The chip designer is now the single biggest driver of global equity returns.

From gaming chips to the top

In October 2022 Nvidia was worth less than a tenth of its current value, a maker of graphics cards for gamers and cryptocurrency miners that had just lived through a sharp downturn in both markets. The launch of ChatGPT the following month changed the story. Training and running large language models requires vast numbers of the parallel processors that Nvidia had spent a decade adapting for data centres, and its software platform, CUDA, had become the default toolkit for the engineers who build them.

The result has been a surge in revenue unlike anything seen at a company of this size. Sales in the latest quarter were more than triple those of a year earlier, and gross margins run above 70 per cent, a level more typical of software than of hardware. Its next generation of chips, Blackwell, unveiled in March, is due to ship later this year, and customers are already queuing for it.

One stock, seven per cent

Nvidia's rise shows how far indices weighted by market value can be bent by one company. It now makes up more than seven per cent of the S&P 500, a level of concentration that adds real risk to the whole market. Passive investors are, in effect, making a large momentum bet that hyperscaler capital spending will keep growing year after year.

Microsoft, Apple and Nvidia are each worth around $3 trillion, and together they account for roughly a fifth of the index. A saver who buys an S&P 500 tracker believing that money is spread across five hundred companies is in practice putting a large slice of it on a single theme. Sector funds concentrate the bet further. Technology funds that cap the weight of any one holding must rebalance each quarter, and when the order among the giants changes, they are forced into large trades that have little to do with any new information about the businesses.

The effect reaches well beyond America. Global indices such as the MSCI World are weighted by market value too, and American technology giants dominate them, so a pension fund in Europe or Japan that buys a world tracker ends up with a sizeable exposure to Nvidia as well. The winners from this concentration are those who held the giants early. The losers are equal-weighted funds and active managers who spread their bets, most of whom have trailed the headline index badly over the past eighteen months, and savers who believed that indexing bought them diversification without further thought.

Where the money comes from

The chain of cash is short. Microsoft, Alphabet, Amazon and Meta are spending tens of billions of dollars a year on data centres, and a large share of that goes to Nvidia. Those companies are funding the spending from their own profits rather than from borrowing, which makes the cycle sturdier than the telecoms build-out of the late 1990s. It still depends on their willingness to keep spending ahead of the revenue that artificial intelligence services will eventually generate. So far that revenue is small relative to the spending, and investors in the hyperscalers have tolerated the gap because their core businesses remain highly profitable. Their patience is the variable on which Nvidia's order book now rests.

That is where the risk lies. If the hyperscalers conclude that they have enough capacity, or that the returns on their spending are too slow to arrive, orders could slow sharply. Several of them are also designing their own chips, such as Google's tensor processing units and Amazon's Trainium, which would reduce their dependence on Nvidia over time. Advanced Micro Devices is selling competing accelerators as well. None of these threatens Nvidia's position this year, but together they cap how long margins this high can last.

The Nifty Fifty problem

Concentrated leadership has happened before, and it has ended badly more than once. In the early 1970s a group of large American growth companies known as the Nifty Fifty, including Xerox, Polaroid and IBM, were treated as "one-decision" stocks to be bought at any price. The 1973–74 bear market cut many of them down by far more than the wider index. At the peak of the dot-com boom in 2000, Cisco briefly became the most valuable company in the world before losing most of its value in the next two years.

The fair objection is that Nvidia, unlike many of those names, is growing earnings almost as fast as its share price. Its valuation relative to expected profits is high but not extreme by the standards of the largest technology companies, because analysts' forecasts have risen alongside the stock. Strong fundamentals do not remove the risk, though. They move it from the valuation to the forecasts, which assume that demand keeps growing at a pace no hardware company has sustained for long.

What would break the momentum

The signals to watch come from the buyers, not the seller. The big cloud companies report quarterly results in late July, and their guidance on capital spending for the rest of the year will say more about Nvidia's prospects than any statement from Nvidia itself. The company reports its own results in late August. Export restrictions on sales of advanced chips to China, which Washington has tightened twice since 2022, are a further risk that could be extended with little notice.

Nvidia's valuation is a vote of confidence in AI hardware demand. It also means that any slowdown in Big Tech's chip spending will hit every index fund holder, and the wider market with it. The first real test arrives with the late-July capital spending guidance; if even one of the four big buyers signals a pause, the stock that did most to lift the index this year is likely to do most to pull it down.