This week, an increasing number of hedge fund founders and strategists appeared pessimistic about the AI trade.
Based on market activity, it may not seem like much, but this week on Wall Street, there were some skepticisms over the sustainability of the artificial intelligence industry.
Reports on possible dangers to the AI narrative were released by a number of well-known analysts. Concerns regarding circular financing in AI and the source of funding for these new data-center commitments were rekindled by recent agreements between Oracle Corp. (ORCL) and OpenAI, as well as OpenAI and Nvidia Corp. (NVDA). For the first time, businesses seemed eager to participate by taking on additional debt: Take a look at Oracle’s earlier this week “unusual” jumbo bond sale.
As a number of AI-related businesses faced pressure, growing skepticism seemed to weigh on equities, prompting key indexes to end the week just short of record territory. Nevertheless, Mark Hackett, chief market strategist at Nationwide, stated that investors’ desire to purchase on dips once more contributed to limiting losses.
The Nasdaq Composite COMP was on course to end the day lower for the fourth consecutive day at one point on Friday. Since April, that would have been the longest run of daily drops. The Nasdaq ended a three-week winning streak even though it ended the session in the green.
“This must eventually affect the bottom line.”
Chief strategist at Interactive Brokers Steve Sosnick talked about the similarities between the surge in AI spending and the haste to install fiber-optic cable that drove the late 1990s telecom bubble. A few well-publicized bankruptcies and scandals marked the conclusion of that preceding phase.
Sosnick emphasized what some have called the dangerously insular character of AI-related investment in commentary that was released earlier this week. Nvidia Corp.’s largest clients are the hyperscalers, which include Microsoft Corp. (MSFT), Amazon.com Inc. (AMZN), Meta Platforms Inc. (META), and Alphabet Inc. (GOOGL). Nvidia is a major client of Super Micro Computer Inc. (SMCI) and CoreWeave Inc. (CRWV).
Over the last two years, Super Micro’s stock has experienced a wild ride and has not yet recovered to its all-time high from early 2024. According to FactSet statistics, CoreWeave’s shares have fallen from their June peak, indicating that the demand surge that accompanied its IPO has subsided.
Although the increase of bandwidth throughout the internet era finally turned a profit, Sosnick pointed out that businesses like Global Crossing filed for bankruptcy protection in 2002 because the funds did not arrive in time to save them.
“We’ve seen immense returns generated from the cash that has been invested, not necessarily in terms of bottom-line results at companies, but the market has seen the benefits,” Sosnick said in a MarketWatch interview. “But at some point, this has to translate into the bottom line.”
An arms race
On Wall Street, Sosnick is not the only person who has questioned the unrelenting spending. In a research released on Wednesday, Michael Cembalest of J.P. Morgan Asset Management outlined the effects of the AI boom on the American economy and markets.
He then outlined several possible roadblocks to the AI rollout, such as the strong demand for power, which is already driving up electricity costs for customers in some countries.
According to Cembalest, since ChatGPT’s November 2022 launch, a collection of 41 artificial intelligence-related stocks had accounted for 75% of the S&P 500’s growth. According to him, they are responsible for 90% of the rise in capital spending and 80% of the growth in corporate earnings.
However, the conversation also took into account new financial dangers. Prior to the Oracle-OpenAI agreement, a few number of businesses, including Microsoft, Meta, Alphabet, and Amazon, were able to finance the majority of the AI expenditure surge with their cash flows. Oracle, however, has broken this trend by borrowing a huge amount in an exceptionally big bond issuance this week. By doing so, it has raised the possibility that the AI data-center build-out will become an arms race, with businesses scrambling to borrow.
Cembalest speculated that Chinese companies like Huawei and SMIC might be able to create their own environment for manufacturing advanced chips, which would weaken Nvidia’s hold on technological supremacy and potentially trigger a selloff. The largest risk facing the top-heavy U.S. equity market, in his opinion, is this.
“Kicking the bear case’s tires”
In reports distributed to clients, a group of Barclays equity analysts stated that while they were still hopeful about AI, they seemed more open to investigating some of the concerns associated with the concept.
Service-provider revenues have been growing at a solid rate, according to the Barclays team. AI-related job listings increased in the first half of the year, and S&P 1,500 companies’ earnings calls also featured more mentions of the technology.
However, the Barclays team stated that “kicking the tires on the bear case” is now more valuable because to the size and speed of the data-center build-out.
What if the lack of power makes further construction more challenging or even impossible? What if newly discovered efficiencies result in surplus data-center capacity, or if AI models cease to advance? That is a major concern now, as it was during the DeepSeek-inspired selloff in January.
Venu Krishna, head of U.S. equity-linked strategies, led the Barclays team that wrote the study. “The lingering fear, then as now, is whether AI is due its ‘dark fiber’ moment, in which exponential efficiencies lead to critical underutilization of infrastructure assets,” the team stated.
The phrase “dark-fiber moment” refers to the telecom bubble that coincided with the dot-com stock bubble in the late 1990s. Fiber-optic cable spending surged due to expectations of skyrocketing internet usage. However, these businesses soon realized that a large portion of their creations would remain dormant, at least initially.
Cisco Systems Inc. (CSCO) has never seen its split-adjusted market value return to its high in the early 2000s, despite riding this trend to briefly become the most valuable business in the world.
The Barclays team also looked into the possibility that debt-driven expenditures would accelerate the build-out, increasing the possibility that investors might be left holding the bag if a large number of data centers turn out to be, in effect, stranded assets. In his report, Cembalest expressed a similar worry.
Some even joined in from the hedge-fund industry. Earlier this week, Ken Griffin of Citadel told CNBC that “there were obviously echoes of the dot-com bubble” in this moment’s AI. During a panel discussion, David Einhorn, the founder of Greenlight Capital, cautioned that unrestrained data-center spending may destroy vast sums of money.
Additionally, the Wall Street Journal questioned whether investing in AI would ever pay off in a piece that appeared in its print edition on Friday.
There are indications that the AI trade is starting to plateau beneath the surface. For over two months, Nvidia, a company that creates the chips required to train AI models, has experienced no increase in its stock price. According to FactSet statistics, Nvidia’s closing price on Friday was $178.19, which was comparable to values from early August. Nvidia was expected to end the week slightly higher, but other semiconductor and Big Tech companies that are directly related to the AI trade were expected to fall.
Technically speaking, earlier this week, chart watchers noticed indications that the AI trade was beginning to appear a little stretched. On Monday, the 14-day relative-strength index of the Global X Artificial Intelligence & Technology ETF AIQ hit 82.41, making it the most overbought since its launch. According to FactSet, its RSI peaked at 80.80 on June 15, 2023, the only previous occasion it was above 80. One common momentum indicator that market technicians employ in their analysis is the RSI.
AI-related doubts are not wholly new. In August, Sam Altman, the CEO of OpenAI, called the AI frenzy a “bubble.” Many of these same worries were voiced more than a year ago by those on Wall Street, most notably Jim Covello at Goldman Sachs Group (GS). However, Sosnick added that Oracle’s choice to use the debt markets to fund its agreement with OpenAI contributed to a resurgence of interest this week.
