The $3 trillion AI datacenter boom is a tale of two markets: a “healthy” boom led by cash-rich tech giants, and a “speculative” bubble financed by risky debt. The “healthy” boom is the $1.4tn in spending covered by the cash flow of “hyperscalers” like Google, Microsoft, Amazon, and Meta. These companies are spending $750bn in the next two years on “general purpose” infrastructure that supports their existing, profitable cloud businesses.
The “speculative” boom is the $1.5tn “funding gap” identified by Morgan Stanley. This “unhealthy” part of the market is being financed by “private credit,” a “shadow banking” sector. Lenders are funding “unproven” projects “without their own customers,” according to Gil Luria of DA Davidson.
This “speculative” side is what worries economists. Alibaba’s chair has warned of a “bubble.” The Uptime Institute notes “many” announced projects “will never be built.” And an MIT study found 95% of AI pilots yield “zero return,” questioning who will pay for this speculative capacity.
The “healthy” boom, rooted in $100bn quarters (Alphabet) and $5tn valuations (Nvidia), is real. But it is “debt-fueled exuberance” in the “speculative” market that poses a “structural risk to the overall global economy.” The $3tn question is whether the “speculative” bubble will burst and drag the “healthy” market down with it.