The $3 trillion projected investment in AI datacenters by 2028 is raising serious questions about financial stability. While tech valuations soar, with Nvidia hitting $5tn, a significant portion of this boom is being financed by risky, opaque debt, prompting warnings of a potential bubble.
Analysts at Morgan Stanley have identified a $1.5 trillion funding gap in the datacenter build-out. This is the amount that the cash-rich “hyperscalers” like Google and Microsoft cannot cover themselves. To plug this hole, developers are turning to private credit, a “shadow banking” sector that has alarmed regulators like the Bank of England.
This is where the risk lies. Gil Luria of DA Davidson distinguishes between the “healthy” investment by Big Tech and the “speculative assets” being funded by this new debt. He warns that lenders, caught in the AI hype, “may not be properly assessing the risks” of backing unproven projects built on assets that depreciate very quickly.
This debt-fueled exuberance is happening despite unclear returns. A hedge fund founder warned datacenters depreciate twice as fast as the revenue they generate. Compounding this, a recent MIT study showed 95% of organizations are getting zero return on their generative AI pilots, questioning the “lofty revenue expectations” the industry is counting on.
If these speculative projects, funded by hundreds of billions in private credit, fail to secure customers and go sour, the fallout could be catastrophic. Analysts warn it could trigger “structural risk to the overall global economy,” turning the AI boom into a widespread financial bust.