We saw yesterday that Musk is financing his hyperscaling by making deals with folks in the Middle East. I will talk about this another time, but sovereigns are an untapped new market for hyperscalar infrastructure providers such as NVidia.
But what about the others? How do they invest in hyperscaling without ruining their balance sheets? Here's one technique:
Meta is planning to raise $29 billion through a mix of $26 billion in debt and $3 billion in equity to rapidly build AI data centers, using special purpose vehicles (SPVs) to keep this debt off its balance sheet. This financial engineering allows Meta to avoid showing the large liabilities directly, even though it retains control over the assets.
https://www.chrispaynephoto.com/
Private equity firms and insurers, seeking higher yields without excessive risk, fund these SPVs, creating a complex web of financing that appears opaque. While this approach benefits all parties in the short term, it carries significant risks. The off-balance-sheet nature of the debt obscures true risk exposure, potentially misleading investors and credit markets.
The massive capital influx encourages overbuilding of AI infrastructure, which is highly capital-intensive and speculative in returns, with only a thin equity cushion to absorb setbacks. Additionally, insurers backing these SPVs face asset-liability mismatches, as their long-term liabilities may not align with the illiquid, concentrated investments in AI data centers. Although this situation is not yet a systemic crisis like the 2008 financial meltdown, it echoes familiar patterns of hidden leverage and mispriced risk.
If AI data center returns falter, the fallout could be substantial, exposing vulnerabilities in the intersection of private credit, insurance capital, and tech infrastructure financing. Also, the attractiveness of Data Center investments (who wouldn't prefer to lend to one Mark Zuckerberg instead of 10000 Small Businesses?) makes it harder for investment to reach other sectors. Manufacturing will certainly be hurt.