Tiny Neocloud Firms Fuel AI Bubble Fears Amidst Debt Concerns
Smaller data center operators, often with roots in crypto mining, face investor jitters and growing debt as demand for AI computing power soars.
A class of lesser-known companies, termed "neoclouds," are increasingly central to the booming artificial intelligence sector, but their substantial debt loads and reliance on major tech clients are fueling concerns about a potential AI bubble.
These firms specialize in providing essential computing power, often financing themselves through long-term lease contracts with major technology companies needing rapid access to hardware. They then leverage these contracts and purchased AI chips as collateral to secure loans. This debt-financed model is common due to the significant upfront costs of the necessary hardware.
Industry leader CoreWeave, though a fraction of the size of tech giants like Microsoft and Meta, has amassed approximately $35.1 billion in debt. Analysts suggest that lenders may place significant weight on the creditworthiness of CoreWeave's large customers when extending credit, rather than solely on the neocloud company itself. This dynamic has led some investors to express unease.
In recent weeks, CoreWeave's stock value has fallen by roughly 25%. Similar declines have been observed in shares of other neocloud companies, including Nebius, IREN, and Applied Digital. Debt investors are also reportedly demanding higher returns from firms in this sector, indicating increased perceived risk.
However, CoreWeave CEO Mike Intrator has publicly dismissed these market reactions as "irrational," asserting confidence in the company's business model and highlighting the strong demand from major technology players for its infrastructure and AI services.
Many neoclouds have origins in the volatile cryptocurrency mining industry. Companies like CoreWeave, IREN, Core Scientific, and Applied Digital initially focused on mining cryptocurrencies such as Ethereum. The physical infrastructure developed for crypto mining proved readily adaptable to the needs of artificial intelligence development when AI experienced a surge in popularity following the release of OpenAI's ChatGPT in 2022.
An exception is Nebius, formerly part of Russia's Yandex NV before it split and rebranded in 2024. The success of these neocloud firms is closely tied to the performance of major AI developers and hyperscalers, who are often their primary customers.
"There is just an enormous amount of spending getting injected into the system," said Melissa Otto, head of Visible Alpha Research at S&P Global, describing the situation as an "arms race." Neoclouds have significantly benefited from this trend. However, Forrester VP and principal analyst Charlie Dai noted that the neocloud model carries "greater financial risk" when market dynamics shift, particularly if major tech clients alter their spending plans.
Recent forecasts from Microsoft and Google that projected lower AI spending than anticipated by Wall Street analysts led to a tumble in the shares of several neocloud firms. The upcoming public offerings of AI powerhouses Anthropic and OpenAI are also seen as pivotal events for the sector.
Should the AI market experience a downturn, neocloud firms would likely "face collateral damage," according to Bradley Tusk of Tusk Ventures.
Furthermore, the ability of neoclouds to secure ongoing financing is complicated by rising interest rates, which make borrowing more expensive. Elon Musk has suggested that this financial pressure could lead to significant consolidation within the sector, with over 200 neoclouds potentially facing closure, particularly the smaller entities.
Despite these concerns, analysts at Yorkville Ives maintain that demand for computing power continues to outstrip supply, with neoclouds filling a crucial gap that larger data center operators cannot address. Nevertheless, recent setbacks, such as Microsoft's cancellation of a deal with British neocloud firm Nscale due to anticipated power supply issues lasting until 2030, and Firmus Australia shelving its IPO plans, add to the unease. These events coincide with public scrutiny of data centers and questions about AI's impact on worker productivity, leading some observers, like analyst Cory Johnson, to express a "reason to worry."