George Noble Warns of an AI Buildout “Doom Loop”

Summary

George Noble’s post relays an argument from Ed Zitron that the AI infrastructure expansion has become a debt-driven feedback loop. The post compares the market to a restaurant whose customers pay with investors’ money while higher activity increases ingredient and loan costs. It says growing AI debt is pushing up Treasury yields, while chip orders increase demand for memory and copper needed for further expansion. According to the post, 70% to 80% of the AI revenue collected by hyperscalers can be traced to OpenAI and Anthropic, which are described as losing billions and still dependent on venture funding. Zitron estimates that large cloud companies receive about $183 billion in AI revenue, compared with an infrastructure buildout that would require trillions of dollars annually to pay for itself. The post argues that valuations must keep rising to create exits, worsening the eventual downside. It also warns that losses could reach households through debt held in pension and life-insurance portfolios. These claims are presented as Zitron’s estimates and analysis, relayed by Noble, rather than as independently verified figures in the post.