Silicon Valley can't hold on anymore, shaking up Wall Street—The "AI arms race" is starting to spread, and so are the risks!
```
The AI "arms race" among tech giants is evolving into a complex financial game.
As annual capital expenditures of tens of billions of dollars become the norm, even Amazon, Google, Meta, Microsoft, and Oracle—companies with cash reserves exceeding $340 billion—have begun to feel unprecedented financial pressure.
They are breaking from the tradition of relying entirely on internal funds to build infrastructure, instead turning to Wall Street for more sophisticated financial solutions. They seek to provide ammunition for this expensive race without harming their own financial stability, but risks are emerging as well.
The “Sweet Burden” of Trillion-Dollar Giants: AI Infrastructure Costs Drive Financing Innovation
In the past, tech giants were accustomed to using their massive internal cash flows to build data centers. But the rise of AI has completely changed the game.
The speed and scale of this race have forced them to look for external capital. Investors and credit rating agencies are closely watching how these tech giants will pay for AI data centers, and whether these massive investments can be converted into new revenues.
To maintain healthy balance sheets while aggressively expanding, tech giants are teaming up with bankers to devise increasingly complex financial strategies. The core objective is one: to move part of the cost and risk off their own balance sheets.
In this context, financial instruments rarely heard of in the tech circle—such as joint ventures, backstop agreements, and syndicated debt—are all being brought to the table.
Three “Financial Plays” to Share Risk
In the exploration of “externalizing” risk and cost, three innovative financial “plays” have surfaced, all centered on cleverly externalizing risks and liabilities.
1. Meta’s “Off-Balance-Sheet” Strategy: Joint Ventures
Meta launched a $29 billion financing plan for its “Hyperion” data center project in Louisiana.
The core structure is a joint venture with investment firm Blue Owl Capital. Blue Owl invests $3 billion in equity, while debt giant Pimco, with Morgan Stanley’s assistance, syndicates the massive $26 billion debt required for the project.
The key to this structure is that Meta will repay the debt in the form of lease payments in the future, thereby moving the entire project off its own balance sheet and controlling debt levels.
2. Oracle’s “Risk Sharing”: Syndicated Loans
As the world’s fourth largest cloud service provider, Oracle recently agreed to become a tenant of a 1.4 GW data center complex under development by Vantage Data Centers, one of the largest such projects in the world.
Given the project’s massive scale, developer Vantage is working with six banks led by JPMorgan Chase, Mitsubishi UFJ Financial Group, and Goldman Sachs to syndicate $22 billion in debt for the project.
This model disperses risk among multiple lenders, reducing the risk exposure of any single institution and making such massive financing possible.
3. Google’s “Ingenious Design”: Backstop Guarantee
Google’s solution is the most complex and ingenious: the backstop guarantee.
In this deal, Google provided a backstop guarantee of up to $3.2 billion for a lease agreement between cloud startup Fluidstack and data center owner TeraWulf, thereby obtaining a 14% stake in TeraWulf.
The ingenuity lies in the guarantee being a contingent liability—it is only triggered if Fluidstack defaults, so Google likely does not need to record it as a current liability.
With Google’s support, TeraWulf last month raised $1 billion through convertible bonds underwritten by Morgan Stanley and Cantor Fitzgerald, more than double its initial financing target.
TeraWulf’s CFO said at a conference last month:
“It’s not easy to get a $2 trillion company, its leadership team, board, and everyone else to agree to a novel concept, but I hope we have provided a roadmap.”
Concerns Amidst the Frenzy: Overheating, Concentration, and Leverage Risks
The enormous financing needs of tech giants coincide with a credit market awash with cash.
Private credit funds holding tens of billions of dollars in investable funds, as well as banks increasingly comfortable with projects that have “investment-grade” tenants, are actively pouring in. The loan-to-cost ratio for data center projects has risen significantly compared to the past.
Jason Tofsky, Global Head of Digital Infrastructure at Goldman Sachs, said that lenders are now willing to provide 80–90% of data center project costs. According to JLL, data center lenders typically provide 65–80% of the total cost of new development projects. Tofsky said:
“There is enough money in the market to fund projects well known to the market. The market can absorb these projects very well.”
However, the capital frenzy is breeding new risks.
First is the risk of market overheating. UBS analysts warned in a report last month that the flood of private credit into the data center space may drive AI development but also “increases the risk of market overheating.”
Second is the risk of high concentration. Data center leases are highly concentrated in a few creditworthy tech giants. This raises concerns: if any of these companies cuts spending due to strategy shifts or suffers a ratings downgrade, the entire ecosystem could be at risk.
Finally, some companies’ leverage risks are already apparent. In July, Moody’s and S&P warned Oracle that as it enters the AI infrastructure building phase, its leverage ratio (currently 4.3x) is much higher than other “hyperscalers.” Unless Oracle brings the ratio below 3.5x, its credit rating could be downgraded.
A Moody’s analyst wrote in a credit report:
“While several other hyperscalers are building AI infrastructure, none have entered this phase with leverage so high and cash flow so negative as Oracle.”
Risk Warning and DisclaimerThe market contains risk, and investments must be made cautiously. This article does not constitute individual investment advice, nor does it take into account the individual investment objectives, financial situation or needs of any user. Users should consider whether any opinions, views, or conclusions in this article are suitable for their particular situation. Investing accordingly is at your own risk. ```