The $3 trillion AI datacenter boom is facing a $1.5 trillion problem. That’s the “funding gap” analysts at Morgan Stanley estimate will be left after cash-rich “hyperscalers” like Google and Microsoft pay their share. How this gap is filled is “raising the alarm” at central banks and worrying economists.
The $1.5tn is expected to be covered by “other sources,” primarily private credit. This “shadow banking” sector, which is less regulated, is “eager to deploy capital into AI.” Mark Zuckerberg’s Meta has already tapped this market for $29bn to fund a datacenter expansion.
This is where the “speculative” bubble risk emerges. Gil Luria of DA Davidson warns that these private lenders “may not be properly assessing the risks” of investing in “a new unproven category.” They are funding projects that may have no customers, built on “very quickly depreciating assets.”
If these “lofty revenue expectations” for AI fail to materialize—a risk highlighted by an MIT study showing 95% of AI pilots have zero return—this $1.5tn in debt could go bad. Luria cautions that this “influx of debt capital… could end up representing structural risk to the overall global economy.”
While headlines focus on Nvidia’s $5tn valuation and Microsoft’s new datacenters, the real story might be the $1.5tn in risky debt building up in the shadows. This “funding gap” could be the fault line that triggers the collapse of the AI boom.
