$120 billion in off-balance-sheet financing! Tech giants team up with Wall Street to play the AI infrastructure game, with risks shifting toward private credit.

$120 billion in off-balance-sheet financing! Tech giants team up with Wall Street to play the AI infrastructure game, with risks shifting toward private credit.

In order to stay ahead in the expensive artificial intelligence (AI) arms race while maintaining flawless financial statements, Silicon Valley tech giants are using complex financial instruments to shift massive infrastructure expenditures off their balance sheets.

According to a June 24 analysis by the Financial Times, tech companies including Meta, xAI, Oracle, and data center operator CoreWeave have used special purpose vehicles (SPVs) to transfer more than $120 billion in data center financing debt to Wall Street investors. This strategy, while protecting the giants’ credit ratings, has also stirred market concerns about opaque risks and potential financial contagion effects.

In this financing frenzy, financial institutions such as Pimco, BlackRock, Apollo, Blue Owl Capital, and banks like JPMorgan have injected at least $120 billion into computing infrastructure projects operating through SPV structures via debt and equity. This innovative structure enables companies like Meta and Oracle to obtain massive funding needed to build AI data centers without significantly increasing their own reported debt. A practice unthinkable 18 years ago, it has now become the industry norm.

However, this wave of financing may hide the actual risks borne by tech groups. Though the SPV structure nominally isolates debt, questions remain about who bears the cost if AI demand falls short of expectations. Industry insiders worry that if AI operators face financial stress in the future, such pressure could unpredictably spread via SPV structures through Wall Street and the private credit market. Morgan Stanley previously anticipated that tech firms’ AI plans would require about $1.5 trillion in external financing, hinting that this financing model may continue to expand in scale.

The “Perfect” Balance Sheet Game

Silicon Valley giants have long been known for strong cash flows and very low debt, which grants them premium credit ratings and investor trust. However, the race for advanced AI computing power has forced these groups to endure unprecedented borrowing pressure. To avoid showing high leverage on their balance sheets—thus protecting credit ratings and beautifying financial indicators—private capital introduced through off-balance-sheet structures has become a preferred path.

Reportedly, a senior executive at a major financing institution stated that, owing to their excellent credit, tech companies can obtain more capital than any other industry. The basic logic of the structure is: Tech companies do not borrow directly; instead, SPVs raise funds to build data centers, and subsequently sign leasing agreements with the companies. In case of default, lenders’ claims are usually limited to the assets under the SPV—data centers, land, and chips—rather than the tech parent company that manages the sites.

The Giants’ Specific Playbook

Meta completed a landmark deal last October. By partnering with New York financial firm Blue Owl Capital to create an SPV called “Beignet Investor,” Meta raised $30 billion for its planned Hyperion facility in Louisiana, including roughly $27 billion in loans from institutions like Pimco, BlackRock, and Apollo. This transaction allowed Meta to effectively borrow $30 billion without displaying any debt on its balance sheet, paving the way for another $30 billion refinancing in the corporate bond market weeks later.

Oracle is also actively using third parties to assemble large debt deals. Led by Larry Ellison, the company has reached multiple agreements with partners such as Crusoe and Blue Owl Capital. Blue Owl and JPMorgan have invested around $13 billion, including $10 billion in debt financing, in an SPV holding Oracle's Texas data center. Additionally, the company arranged two sizable bundled financings—about $38 billion and $18 billion each—for various data center projects in Texas, Wisconsin, and New Mexico. In these cases, Oracle has agreed to lease facilities from the SPVs.

Furthermore, Elon Musk’s startup xAI also used a similar structure as part of its $20 billion financing, including up to $12.5 billion in debt. The SPV will use these funds to purchase Nvidia GPUs and lease them to xAI.

Concerns in the Private Credit Market

As private capital investors eagerly join the AI boom, “project finance” transactions focused on long-term infrastructure financing are surging. According to UBS data, by early 2025, tech companies will have borrowed about $450 billion from private equity funds, up $100 billion year-over-year. This year, roughly $125 billion has flowed into project finance deals similar to the Meta and Blue Owl arrangement.

This trend has fueled market worries about the rapidly ballooning $1.7 trillion private credit industry: soaring asset valuations, lack of liquidity, and borrower concentration risk. A banker close to data center financing deals noted that risky loans and potential credit risk have already built up in private credit. Because the AI data center boom mainly depends on a handful of clients (OpenAI has made over $1.4 trillion in long-term computing commitments), if a major tenant encounters issues, lenders to multiple data centers may face shared risk exposure.

Investors also face uncertainty concerning power supply, AI regulation changes, and technological shifts rendering hardware obsolete. Wall Street is even attempting to securitize AI debt by packaging loans for broader sale to investors such as asset managers and pension funds—these deals currently total tens of billions of dollars.

Risk Exposure and Divergent Strategies

Although financing structures aim to isolate risk, in many cases, if dwindling AI demand hurts facility values, the ultimate financial risk often still falls on the tech company leasing the sites. For instance, in the “Beignet Investor” case, Meta owns 20% of the SPV and has provided “residual value guarantees” to other investors. This means if the data center’s value falls below a certain threshold and Meta decides not to renew the lease, it must compensate SPV investors.

UBS public and private credit strategist Matthew Mish pointed out that while most investors view taking on "hyperscaler" risk as desirable, SPV financing actually increases tech firms’ unreported liabilities, casting doubt on the true strength of their overall credit quality.

Not all giants have joined this off-balance-sheet financing trend. Google, Microsoft, and Amazon—already seasoned operators in data centers before the AI boom—still mainly finance new builds using cash or direct bond issuance, and have yet to disclose any major SPV financing plans. This divergence in strategy reflects different approaches to risk management among players confronted with costly AI bets.

Risk Warning and DisclaimerThe market entails risks, and investments should be made cautiously. This article does not constitute personal investment advice and does not take into account the individual investment goals, financial situation, or needs of any user. Users should consider whether any opinions, views, or conclusions herein fit their specific circumstances. Investments made based on this information are at your own risk.