A recent proposal from Donald Trump to have tech giants fund their own energy sources for data centers is an important development. The stated goal is to prevent the massive energy consumption of AI from causing electricity rates to spike for the average citizen. While the intention is sound, I believe there’s a significant blind spot in this approach.
Building this infrastructure requires immense capital. To avoid a direct hit to their balance sheets, tech giants are turning to complex financial solutions. We see this in the roughly $27 billion financing agreement between Meta and Blue Owl for the Hyperion campus, and in the Global AI Infrastructure Investment Partnership (GAIIP), which is targeting an initial $30 billion for the same purpose.
The funding for these deals often comes from instruments like Data Center Asset-Backed Securities (ABS) and Commercial Mortgage-Backed Securities (CMBS). In 2025 alone, issuance in this sector exceeded $20 billion. Critically, the buyers of this debt include large institutional investors, many of whom manage public pension savings.
This creates a hidden risk. A rapid technological shift, such as a move away from massive centralized models, could lead tech giants to exit their contracts early. The resulting financial loss could roll directly onto pension funds, ironically hitting the same citizens the policy aimed to protect.
While there’s nothing inherently wrong with these financial mechanisms, and large tech companies have alternative uses for these facilities, it's a complex ecosystem. We must be vigilant and ensure the AI hype doesn't quietly create the conditions for another 2008-style financial crisis.