This post was originally published on this site
https://content.fortune.com/wp-content/uploads/2023/07/GettyImages-812897322-e1689273661608.jpg?w=2048The large language models widely known as artificial intelligence could lead to an $800 billion opportunity, some analysts say, with companies battling it out Game of Thrones-style to carve out their position in the market. That’s why companies diving into the field are making such big bets: Microsoft recently invested $10 billion in the ChatGPT-maker OpenAI; Databricks bought the machine learning startup MosiacML for $1.3 billion; and the niche legal A.I. platform Casetext sold for $650 million to Thomson Reuters.
But despite the current buzz around A.I., the stocks of the major players in the new technology are overvalued, says Dhaval Joshi, chief strategist at the investment research firm BCA Research, in a report released Wednesday.
“This year’s near-vertical rally in the A.I. seven is premised on a hope,” Joshi writes, using a new nomenclature for the stocks of Meta, Amazon, Alphabet, Microsoft, Nvidia, Apple, and Tesla. “The hope is that the latest iteration of the internet boom—this time led by generative A.I.—will supercharge technology sector profits.”
He argues, however, that without a clear pathway to revenue, and eventually profits, that hope risks turning the A.I. boom into a bubble akin to the dot com bubble in the 1990s.
To avoid a similar scenario with A.I., companies would need to establish a “moat” that keeps out competitors while monetizing a new innovation, as tech companies such as Alphabet, Meta, and Amazon succeeded in doing only during the 2000s, Joshi says. He calls these companies the “Web 2.0 oligopolies.” They were able to establish a moat using the so-called network effect in which the value of a product increases as more people use it. This effect allowed Google, Meta, and Amazon to build ever-growing user bases in search, social media, and e-commerce, respectively, and then use that market dominance to squeeze out smaller competitors.
“The crucial lesson is that there is no guarantee that an innovative technology can supercharge the profits of its creator,” Joshi writes. “To supercharge its profits, the creator must have a ‘moat’ to prevent its profits being competed away.”
Without A.I.’s version of a moat, companies might struggle to monetize the technology and fail to live up to their current valuations, Joshi says. So far, Joshi is skeptical that the consumer-facing versions of generative A.I., such as ChatGPT or Bard, will distinguish themselves from one another enough to be commercially viable.
“There’s no reason why any particular A.I. should be the one that everyone goes to,” he tells Fortune. “There’s no reason a priori why Bard will be inferior to ChatGPT or any other competitor.”
A.I. companies also need to factor in the price of costly parts, especially computing chips. Meanwhile chipmakers such as Nvidia—whose $1.1 trillion market cap is roughly 41 times its current revenue—risk being unable to live up to their current valuations if they can’t find successful A.I. companies to sell their products to.
Those products are expensive, Joshi points out: “At the moment the biggest companies can afford it, but you go down the pecking order and people can’t. So, you either reduce prices to increase demand, or you’re going to hit some sort of speed limit. That’s why I think some of these growth expectations seem a bit too rich at the moment.”
We may soon know whether Joshi’s wariness is justified. Either a drop in prices or a slowdown in Nvidia’s sales, coupled with ChatGPT’s first month-over-month declines in site traffic, could signal the “cold water moment” in which A.I. stocks start to rightsize, he says.
“Everyone’s riding this scene,” Joshi says. “But what we’re going to discover over the next few months or years is if the first stab or first guess of who is going to be riding the scene is correct. Probably not.”
