OVER three decades on Wall Street, Jim Covello has learned how painful it can be to bet against an inflating tech stock bubble. The market has a way of minting riches, month after month, even after it’s clear the latest breakthroughs are not playing out quite as expected.
It happened with dot-com companies in the late 1990s and more recently with cryptocurrencies. And Covello, the head of equity research at Goldman Sachs Group, said it will likely happen with artificial intelligence (AI), too, making it dangerous, if not outright foolish, to start wagering against the likes of Nvidia.
And yet, he has no doubt that the reckoning is coming. It might not be this year or even next year, for that matter, but at some point, he said, it’s happening. As he sees it, the hundreds of billions of US dollars companies are ploughing into AI will not set off the next economic revolution – or even rival the benefits of the smartphone and the Internet. And when that becomes clear, all the stocks that have surged on the back of its promise will come sliding down, too.
“Most technology transitions in history, particularly the ones that have been transformational, have seen us replace very expensive solutions with very cheap solutions,” said Covello, who first made a name for himself at Goldman as a tech stock analyst. “Potentially replacing jobs with tremendously costly technology is basically the polar opposite.”
Covello is emerging as a leader of a small but growing cohort of market watchers who are casting doubt on a crucial tenet of the rally that’s added nearly US$16 trillion to the S&P 500 since late 2022: The idea that the dazzling power of large-language models will usher in the next great stage of capitalism, one where corporate profits will boom as more and more work is handed over to intelligent machines, boosting efficiencies and accelerating growth.
There are plenty of believers. Jamie Dimon, JPMorgan Chase’s chief executive officer, has said he’s convinced that AI will unleash extraordinary changes, potentially as transformative as those brought by the printing press, the steam engine and electricity. State Street’s chief investment strategist, Michael Arone, said it’s bringing a “prolonged and unprecedented productivity miracle”. Even inside Covello’s own firm, senior global economist Joseph Briggs estimates AI will ultimately automate a quarter of all work tasks and provide a boost to the pace of economic growth.
BT in your inbox
Start and end each day with the latest news stories and analyses delivered straight to your inbox.
Such speculation, in turn, has set off a very concrete boom as the world’s biggest tech companies invest heavily in a bid to extend their dominance to the newest sphere. That’s been a bounty for companies such as Nvidia, Broadcom and Super Micro Computer that provide the hardware needed to power AI models. Even utilities have seen sales jump amid soaring demand from electricity thirsty data centres.
The sceptics said the problem is that the commercial hopes for the technology may be significantly overblown, creating the risk of a stock-market correction if the tech Goliaths rethink their massive investments.
‘The last sucker’
David Bahnsen, founder and chief investment officer at the Bahnsen Group, has been preparing for that scenario. He has avoided Nvidia and other large tech stocks, anticipating what he sees as a potential “disaster”.
“The way we make money is to not be holding the bag when the last sucker places that trade on Cisco in March of 2000,” he said, referring to the precipitous drop that the hardware company’s shares suffered after the dot-com bubble burst. “A lot of people are going to lose money if they don’t trade out.”
So far, there’s been few signs of that. While tech stocks tumbled on Wednesday on worries chipmakers will be pulled more deeply into a US-China trade war, they are still holding near record highs. Nearly half of the S&P 500’s advance since it bottomed out in October 2022 have been concentrated in just six stocks: Apple, Microsoft, Nvidia, Alphabet, Amazon.com and Meta Platforms.
Nvidia, which alone has added nearly US$2 trillion in market value this year, remains one of the most popular stocks on Wall Street: Sixty four of the analysts who follow the chipmaker are still advising clients to buy it – even after a nearly 140 per cent jump this year. Only one says to sell.
The payoff from all that AI investment so far, though, has been relatively modest.
Microsoft, Alphabet, Amazon, and Meta have collectively pumped more than US$150 billion into capital expenditures in the last four quarters, with much of that for computing capacity to train their own large-language models and run workloads for customers.
Microsoft, which has been infusing OpenAI’s technology across its product line, said in April that AI services contributed seven percentage points to the 31 per cent expansion of sales from Azure and other cloud services in its fiscal third quarter, without giving the US dollar amount.
Amazon, with projected sales of more than US$600 billion this year, has said only that its AI business has a “multi-billion dollar revenue run rate”. On Alphabet’s first quarter earnings call, Chief Financial Officer Ruth Porat acknowledged “an increasing contribution from AI ” to Google Cloud revenue.
To bulls like Adam Gold, whose largest position is Nvidia shares, the focus on some of those numbers is premature. The Katam Hill chief investment officer notes that companies such as Meta – which does not charge users – are already seeing sales growth by using AI to improve algorithms for ad placement and engagement.
“These are the smartest, best capital allocators in the world,” he said. “They are building these data centres to drive profitability in out years.”
For some of the customers of the cloud-computing giants, the gains have been less evident. In May, Salesforce shares tumbled after the software company projected the slowest quarterly sales growth in its history despite long touting the potential for AI to boost sales. According to a survey conducted by San Francisco-based Lucidworks, fewer than half of companies investing in AI have yet to see a significant return.
‘Bubbles take time’
Covello is doubtful that most ever will. He has followed the tech industry’s ups and downs since joining Goldman in 2000 after stints with SG Cowen and Smith Barney. He specialised in semiconductor equipment companies, wracking up awards year after year as the sector’s top analyst, before being promoted in 2015 to the firm’s director of Americas equity research.
To get an adequate return on the approximately US$1 trillion of AI infrastructure investment he expects to see over the next several years, companies would need to be able to use it to solve increasingly complex tasks.
As he sees it, AI has shown the promise of making some jobs like coding more efficient, but not nearly enough to justify the expense.
If significant uses do not start to become apparent in the next year and a half, Covello said, the stock-market tide will turn. Yet he does not think it’s nearing that stage yet, with the continued buildout likely to keep driving investors into stocks such as Nvidia.
“One of the most important lessons I have learned over the past three decades,” he said. “Is that bubbles can take a long time to burst.” BLOOMBERG