Oracle, Meta, xAI, and CoreWeave are channeling over $120 billion into AI data center infrastructure by leveraging Wall Street financial mechanisms, particularly special purpose vehicles (SPVs), to manage the mounting costs associated with artificial intelligence development. This strategic financing allows these tech giants to secure vital computing resources while keeping significant debts off their core financial statements. According to the Financial Times, major investment firms, including Pimco, BlackRock, and Apollo, as well as large U.S. banks such as JPMorgan, are backing these initiatives.
The structure of these SPVs enables the tech companies to access necessary capital while mitigating their balance sheet exposure. The funds raised are allocated to acquire land, structures, and essential computing chips that the companies then lease. This innovative financing model marks a significant shift from the traditional reliance on cash reserves and modest debt levels that once characterized Silicon Valley.
Meta, for instance, set a significant precedent in October with a $30 billion private credit agreement for its planned Hyperion data center in Louisiana. This project is managed through an SPV named Beignet Investor, established with Blue Owl Capital. The investment vehicle successfully secured approximately $27 billion in loans from prominent investors, coupled with $3 billion in equity from Blue Owl, allowing Meta to keep this debt hidden from its balance sheet. This maneuver also paved the way for the company to raise an additional $30 billion in corporate bonds the following month.
Oracle has adopted a similar strategy, leasing computing power to OpenAI while collaborating with various builders and financiers, including Crusoe and Vantage. The company’s approach involves multiple data centers, each funded through separate SPVs. A notable arrangement for a facility in Abilene, Texas, received $13 billion from Blue Owl and JPMorgan, which included $10 billion in debt. Other financing strategies comprise a $38 billion debt package for locations in Texas and Wisconsin, alongside an $18 billion loan for a facility in New Mexico. In all instances, Oracle leases these facilities, with lenders retaining claims on the assets.
As these SPVs proliferate, the risk associated with such financial structures is becoming more dispersed across Wall Street. Investors often perceive the primary risk still resides with the tech companies leasing the data center sites. This dynamic is evident in Meta’s Beignet Investor deal, where the company owns a 20% stake in the SPV and has provided a residual value guarantee. This guarantees repayment to investors if the data center’s value diminishes below an agreed threshold upon lease expiration, should Meta choose not to renew.
Elon Musk’s xAI is also following this trend, working to raise $20 billion, which includes up to $12.5 billion in debt. A newly formed SPV will utilize these funds to acquire Nvidia graphics processing units and lease them back to xAI, mirroring the financing strategies of its counterparts. CoreWeave is pursuing a similar path; in March, it established an SPV to fulfill an $11.9 billion contract to supply computing power to OpenAI, later borrowing $2.6 billion to support these obligations.
A burgeoning interest in private capital persists, with UBS reporting that tech companies had borrowed around $450 billion from private funds by early 2025, an increase of $100 billion from the previous year. Approximately $125 billion has been funneled into project finance deals in just this year alone. The construction of data centers now heavily relies on the expansive $1.7 trillion private credit market, although concerns about valuation, illiquidity, and borrower concentration are intensifying.
As more companies adopt these financial structures, the risks associated with such arrangements are escalating. Notably, OpenAI has committed over $1.4 trillion in long-term computing contracts across the industry. Meanwhile, tech giants like Google, Microsoft, and Amazon have so far opted against SPVs, instead funding their data centers through cash reserves and traditional bond issues. This divergence in financing strategies may shape the future landscape of AI infrastructure investment.
See also
BofA Reveals 6 Small-Cap Stocks Set to Thrive in $1.2 Trillion AI Market by 2030
BILL.com Targets 95% Market Gap with AI Financial Automation, Processing $89B Quarterly
Sitharaman Launches Cyient AI & Future Skills Hub to Empower Rural Communities in India
Taiwan Semiconductor Manufacturing Set for Explosive Growth Amid AI Chip Demand
AI Transforms Financial Compliance: Real-Time Monitoring Redefines Risk Assessment
