Are the Nvidia skeptics right?

Unlock Editor's Digest without spending a dime

FT editor Roula Khalaf selects her favourite stories on this weekly newsletter.

Good morning. Bill Hwang, whose hedge fund Archegos blew up spectacularly a couple of years ago, has been found guilty of fraud. I still have the identical query about Hwang as I at all times do. How could he have thought his crazy plan would generate profits ultimately? Send your theories: robert.armstrong@ft.com.

The Nvidia skeptics

Before we discuss whether it continues to be advisable to own Nvidia stock, let's take a fast take a look at what the expectations for the corporate are. Here is an image of historical and expected earnings per share. It also shows what the expectations for earnings per share were a 12 months ago:

Nvidia's revenue is anticipated to double to $120 billion in the present fiscal 12 months. That's no exaggeration: In May, the corporate said it expected revenue of $28 billion in the present quarter (the second of the fiscal 12 months). Even more notable, analysts expect revenue to greater than double again in the next 4 years, reaching $245 billion within the fiscal 12 months ending in January 2029. For comparison, total revenue from all semiconductors was $533 billion in 2023, in accordance with Gartner.

In other words, the large expansion of AI hardware is much from reaching its peak. And analysts don’t expect competition to extend either: operating margins are expected to exceed 50 percent by 2020. Hence the breathtaking increase in profits shown above.

I'm undecided how much we will read into analysts' forecasts for greater than a couple of years into the long run apart from loud, enthusiastic tones. But Nvidia's stock has risen from 25 times this 12 months's earnings to 45 times earnings prior to now six months, suggesting that investors either imagine in a future just like what analysts are predicting or are playing a game that has nothing to do with earnings expectations in any respect.

This burgeoning enthusiasm is the backdrop against which we must evaluate the arguments of Nvidia skeptics. Loyal readers of Unhedged have already seen two examples of this rare breed. Here's NYU professor Aswath Damodaran from March explaining why he sold a big portion of his Nvidia position (which he made a fortune from):

With Nvidia's ($85), it seems to me that the value is ahead of (history). I did a reverse engineering evaluation to find out how big the AI ​​market and the way big Nvidia's share would need to be to justify the value, and it got here out to ($45). And I calculated that the AI ​​chip market would need to be about $500 billion and Nvidia would need to have 80 percent of the market to essentially break even.

The stock price is currently at $135, so you may imagine what the implied value of the AI ​​chip market looks like immediately.

Another type of skepticism says that if Nvidia is to grow as much as its price implies, its customers' AI investment plans should be huge. And while these customers' plans are big, they aren't large enough. Either firms like Microsoft, Google, and Meta have to spend greater than currently expected, or Nvidia must earn lower than expected. Charles Cara of Absolute Strategy Research raised this point, as we discussed a couple of weeks ago.

A multi-part skeptical argument about AI technology and thus Nvidia was recently made by Jim Covello, head of worldwide equity research at Goldman Sachs and long-time technology analyst. It goes something like this:

1. Building and running AI capabilities is dear – far more expensive than the technologies and business processes they’re designed to exchange. This is in contrast to previous technological innovations, similar to the web, which were cheaper than what they replaced.

2. AI won't get cheaper as more people use it, as many other technologies do, unless the marketplace for the underlying hardware becomes highly competitive, and which may not occur. Nvidia's position in AI chips is like ASML's in high-end chip lithography, which has proven to be almost unassailable.

3. People overestimate the capabilities of AI today. For example, it is just not good at providing easy but accurate summaries of complex information.

4. The big use case for AI has not yet been specified. For previous technologies just like the smartphone, the last word use cases were already specified – normally accurately – when the technology was still in its infancy. AI could lead on to process efficiencies in areas like software programming, but no revolutionary application has yet been found and even proposed.

There's so much to agree or disagree with here (especially: is number 3 really right?), but to me, the interesting thing is what Covello says next. He believes that the pressures of the hype cycle will force firms to proceed investing in AI, and that it's subsequently smart to remain invested in the businesses that can profit from those investments—chipmakers, utilities, etc.—even when those firms seem expensive.

The end of the cycle will only come when there aren’t any lucrative AI applications until the following economic downturn. In an environment of declining profits, investors' tolerance for expensive, low-return experiments will wane and the AI ​​boom will end. Many people describe Nvidia as selling pickaxes and shovels for the AI ​​gold rush. That's a very good metaphor. But if nobody discovers gold before the following recession, the boom will end abruptly.

I find Covello's argument compelling. That's partly because one area where AI seems to have a number of applications is investing, which is ultimately about finding the signal in reams of noisy information. But the small group of quantitative finance experts I consult with about this don't seem particularly impressed. Several of them have said that AI is solely doing what they've been doing for a while, just with more processing power. But Nvidia hasn't priced AI so high that it's an evolutionary technology. The price is for a revolution.

A great read

Puppet people.

FT Unhedged Podcast

Can't get enough of Unhedged? Listen our recent podcastto immerse yourself in the newest market news and financial headlines for quarter-hour twice every week. Read previous editions of the newsletter Here.