AI bubble fears intensify: new cloud companies face debt and stock price pressure

Will the AI bubble burst? These "neocloud" companies are heightening market concerns

Image source, ReutersArticle Information

Author, Lily Jamali

A once little-known corner of the artificial intelligence (AI) ecosystem has now become a market focus, and investors are divided on its prospects.

Company names such as CoreWeave and Nebius are familiar to people in the tech circles of San Francisco and Silicon Valley.

These data center companies are known as "neocloud" firms, operating large data centers filled with expensive computer equipment and leasing computing power to major AI developers such as Anthropic, OpenAI and Meta.

Now they have become one of the most fiercely debated topics in Silicon Valley.

Many question whether neocloud companies' finances are sound, and what their recent setbacks reflect about the current state of the AI economy.

Analysts at business and technology research firm Gartner predict that by 2030, 80% of neocloud companies will "cease to operate independently." On the other hand, experts at Forrester expect that at least three neocloud companies will be acquired in the near term because their "financial condition is becoming unstable."

US Treasury Secretary Scott Bessent tried last Thursday (October 29) to play down concerns about neocloud companies. In an interview with US business channel CNBC, he said these companies are only "a small part of the overall ecosystem."

Debt concerns

Cost-of-living protests in Germany turn violent, shops looted

Mbappé apologizes under Chinese pressure, but many still don't know what he actually said

Behind the "empty chairs" at the CPC's Fifth Plenum: how normalized high-pressure anti-corruption is changing Chinese officialdom

Why are young people across Europe taking to the streets? The shared demands and fears of violence behind it

English text is being prepared.

Cory Johnson, an analyst at IPO research firm The Quiet Period, explained that these companies can then use the relevant contracts and the AI chips they have purchased as collateral to borrow.

He noted that debt is quite common among neocloud companies because the relevant hardware equipment requires upfront payment.

Take industry leader CoreWeave as an example: its scale is far smaller than some of the large tech companies that use its services.

But Johnson believes that when providing loans, lenders sometimes place more weight on the creditworthiness of CoreWeave's customers such as Microsoft and Meta than on CoreWeave's own financial condition.

He said: "This helps explain how CoreWeave accumulated $35.1 billion (233.7 billion yuan; NT$1.15 trillion) in debt."

Some uneasy investors appear to be starting to change their views.

Over the past month, CoreWeave's share price has fallen by a quarter, and it is not an isolated case.

Other neocloud companies, including Nebius, IREN and Applied Digital, have seen roughly similar or even larger share price declines over the same period.

Bond investors also appear rattled and have begun demanding higher returns from several companies in the sector.

CoreWeave CEO Mike Intrator said the market's current reaction is "irrational" and expressed full confidence in the company's business model.

Some of the world's most influential companies are lining up to use CoreWeave's infrastructure and other AI services.

Origins in cryptocurrency

Some neocloud companies originated in the highly volatile cryptocurrency mining industry, producing cryptocurrencies such as Bitcoin.

The infrastructure needed for mining has much in common with the equipment needed to develop AI.

CoreWeave was founded in 2017 and initially focused on Ethereum mining.

IREN, Core Scientific and Applied Digital also started out in cryptocurrency mining.

When OpenAI launched ChatGPT in 2022 and sparked the AI boom, these companies' existing infrastructure gave them a rapid advantage.

Nebius is somewhat different. The company was formerly Yandex NV, the parent company of Yandex, often called "Russia's Google," and completed its spin-off and renaming in 2024.

Today, the success or failure of these companies depends largely on major AI developers and the so-called hyperscalers. The latter are larger data center operators, and neocloud companies often regard them as important customers.

Melissa Otto, head of research at S&P Global's Visible Alpha, said: "There is currently a huge amount of money pouring into the entire system."

"This is an arms race."

And neocloud companies are among the beneficiaries.

However, Charlie Dai, vice president and principal analyst at Forrester, said the cost of this business model is that companies will face "higher financial risk" once conditions change.

This includes any adjustments to spending plans by large tech customers.

Just in the past week, after Microsoft and Google forecast future AI investment levels below Wall Street analysts' expectations, shares of several neocloud companies fell in response.

In addition, the market is also closely watching the development of Anthropic and OpenAI. Both companies are expected to sell shares publicly in the coming months.

Bradley Tusk, founder of venture capital firm Tusk Ventures, said that if AI ultimately proves to be a bubble and the bubble bursts: "Neocloud companies will certainly be hit in the fallout."

"Guppy-level mass death"

The future of neocloud companies also depends on whether they can continue borrowing, and when interest rates rise, financing costs will rise accordingly.

Elon Musk recently posted on a social platform that this challenge could lead to more than 200 neocloud companies facing "guppy-level death and consolidation," with the smallest companies most vulnerable.

He added: "Some companies will be able to survive, but not many."

For now, Yorkville Ives analysts say demand for computing power still exceeds supply, and neocloud companies are filling an important gap that large data center operators cannot meet.

However, in the eyes of some, the recent volatility in the neocloud industry is worrying.

Microsoft recently canceled a cooperation agreement with British neocloud company Nscale due to power supply issues, and the relevant power constraints may last until 2030.

Australian company Firmus has just shelved its IPO plans.

These developments come amid rising public backlash against data centers, and the market is also beginning to question how much practical value AI can actually bring to employee productivity.

Johnson said: "There is indeed reason for concern."

This article was originally written in English. We used AI to assist with translation, and it was reviewed by a BBC journalist before publication. Learn more about how we use AI.