"the AI datacenters to be built in 2025 will suffer $40 billion of annual depreciation, while generating somewhere between $15 and $20 billion of revenue" [$40 billion]
The explicit gap between $40B depreciation and $15-20B revenue directly quantifies the structural mismatch between capex commitment and monetizable demand that defines the overbuild dynamic the signal describes.
Reasoning from this article
The article frames the AI datacenter buildout as a structural repeat of the dot-com fiber overbuild (Global Crossing) and shale capital cycles — cases where the underlying technology proved real but investors were destroyed by capital misallocation. The $400B 2025 datacenter spend requiring ~$480B in revenue just to hit a dilutive 20% ROIC, against current revenues of $15-20B, generalizes to a broader pattern: transformative infrastructure attracts capital far in excess of near-term or even medium-term monetization capacity, and the equity holders funding the buildout absorb the losses when the math becomes undeniable.