Oracle, Meta, xAI, and CoreWeave Raise Funds Through SPVs
There are speculations that four major U.S. big tech companies have used special purpose vehicles (SPVs) to conceal as much as $120 billion (approximately 173 trillion won) in artificial intelligence (AI)-related investment debt from their financial statements. Concerns are mounting over the financial risks stemming from such massive AI investments.
The Financial Times (FT) reported on December 24 (local time) that it reached this estimate after conducting its own analysis of Oracle, Meta Platforms (Meta), xAI, and CoreWeave.
These companies created SPVs to raise funds for AI infrastructure investments, with Wall Street financial firms such as PIMCO, BlackRock, Apollo, Blue Owl, and JPMorgan providing capital by purchasing stocks or bonds issued by these SPVs.
The funds raised in this way flow through the SPVs and do not appear on the big tech companies' financial statements. Since these are not recorded as liabilities, the companies can avoid credit rating downgrades, but FT pointed out that this also allows them to conceal the risks associated with AI investments. The outlet also noted that the SPV structure could increase the risk that, should AI operators face financial difficulties in the future, the resulting crisis could spread across Wall Street in unpredictable ways.
Oracle is cited as the company with the largest amount of such "off-balance sheet debt." FT estimated that Oracle borrowed $66 billion in AI-related funds through SPVs.
Oracle is collaborating with Crusoe, Blue Owl, Vantage, and Related Digital to build data centers. Each SPV will own these data centers, and Oracle has agreed to lease the facilities from the SPVs.
If a default occurs, creditors can exercise rights only over tangible assets such as the data center sites, equipment, and internal chips. Oracle, as the operator of the data centers, would not be held responsible for repayment.
Meta established SPVs called Blue Owl and Benue Investors to raise $30 billion for data centers. As a result, Meta was able to borrow $30 billion without the debt appearing on its financial statements, and a few weeks later, it was able to raise an additional $30 billion in the corporate bond market.
As the capital required to build AI infrastructure surges and big tech companies' cash reserves reach their limits, the practice of raising off-balance sheet debt through SPVs is spreading further. Most data center investors believe that if the value of the facilities declines due to reduced demand for AI services, the financial risk lies with the technology companies leasing those facilities.
Big tech data center construction relies heavily on the private lending market, which has rapidly grown to a scale of $1.7 trillion. However, concerns are rising due to the sharp increase in asset values, lack of liquidity, and borrower concentration.
The current AI data center boom is heavily dependent on a small group of clients. This means that if a major tenant company falters, lenders who have financed multiple data centers could be exposed to the same risks. Additional risks include power shortages, changes in AI regulations, and technological shifts.
An executive at a bank familiar with data center financing transactions stated, "The private lending market already contains risky loans and potential credit risks. The uncertainty of AI investments and the risk of defaults in private lending are closely intertwined, and over the next few years, we are likely to see a very interesting yet concerning phase unfold."
FT also pointed out that if multiple AI companies raise funds through SPVs, a lack of transparency could cause market shocks to spread simultaneously across private lending funds.
Meanwhile, not all big tech companies are following this trend. According to FT, Google, Microsoft (MS), and Amazon are choosing to use internal cash or borrow money directly to expand their data centers.
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
