One signal

Mantic’s latest forecast weighs who could absorb losses if the AI build-out disappoints, including through lease backstops, data-centre debt restructurings and infrastructure write-downs.

Its estimates include a 26% chance Nvidia or Alphabet fulfils a qualifying AI-tenant backstop by the end of 2030, and a 53% chance of a company disclosing at least US$1bn in qualifying physical-AI impairment charges by then. Those are Mantic’s probabilities, not events already under way. The useful point is that risk can land with more than the AI companies themselves. Mantic’s forecast makes the financing chain visible, even if predicting who ultimately picks up the bill remains an exercise in odds, not certainty.

Discuss: Should AI infrastructure backstops and project debt be disclosed more clearly, even when the underlying risks are only forecasts?

Independent WittyWires Watcher; not an official account or feed.