Boom, Bubble, or Both? Inside the $3Tn AI Datacenter Paradox

The $3 trillion AI datacenter boom is a paradox: it is simultaneously a “healthy” technological revolution and a “speculative” bubble, with the “exuberance” of one feeding the other.
The “healthy” boom is the $1.4tn being spent by “hyperscalers” like Google and Microsoft. Backed by $100bn quarters and $4tn valuations, they are building “general purpose” datacenters in places like Newport, Wales, creating “generational employment opportunities.”
The “speculative” bubble is the $1.5tn “funding gap” being filled by “private credit.” This “shadow banking” debt is financing “unproven” projects “without their own customers.” Alibaba’s chair warns this is a “bubble,” and the Uptime Institute says “many” of these projects “will never be built.”
This “exuberance” is running headlong into “zero return” reality. An MIT study found 95% of AI pilots are failing, questioning the “lofty revenue expectations” needed to pay off the $1.S 5tn in speculative debt.
The $3tn market is not one thing. It’s a “healthy” core surrounded by a “debt-fueled” frenzy. The “structural risk” is that when the “speculative” bubble bursts, it will “backfire” and damage the “healthy” revolution at its center.