Investor's Business Daily
Zitron correctly spots the AI casino’s absurdity: hyperscalers fund loss-making startups, startups buy hyperscaler compute, and Wall Street applauds the revenue loop it just financed. But the interview stops at “the bubble will burst” and never asks the DT question—what happens when the underlying tools keep eliminating paid work even after the financiers stop clapping.
Zitron argues that reported returns on AI spending are largely circular. Microsoft, Google, and Amazon derive much of their AI-related revenue from OpenAI and Anthropic, while those companies depend on funding and compute supplied by the same ecosystem. He contrasts relatively small, loss-making AI revenues with hundreds of billions or trillions in capital expenditure, arguing that stock prices, run-rate figures, memoranda of understanding, and cloud growth are being treated as proof of demand without demonstrating durable profitability. He focuses particularly on leverage in data-center infrastructure, citing CoreWeave, planned capacity, and the dependence on continually available credit. He predicts that the first major break will occur when debt becomes harder or more expensive to raise, leading to canceled projects and losses for pension funds, insurers, local communities, and retail investors. He also questions claims about user productivity and product utility, arguing that current AI tools remain unreliable, difficult to measure, and heavily subsidized.
Relative to the Discontinuity Thesis, the interview partially aligns with its rentier analysis. Zitron identifies capital recycling, financial engineering, speculative valuations, and the possibility that losses will ultimately be imposed on pension funds, communities, and ordinary investors. He rejects the comforting claim that current spending already proves a successful transition and explicitly describes the AI build-out as structurally dependent on easy credit and hype. However, the central labour proposition is absent. He does not assess whether AI reduces aggregate demand for human labour, breaks the wage-consumption circuit, or structurally excludes workers. His bubble narrative implies that financial collapse is the decisive correction, but a credit crash would not restore displaced labour demand; nor does weak vendor profitability prove that the technology cannot substitute for paid work. The lack of a comforting solution keeps the score relatively low, but the omission of labour discontinuity prevents a lucid classification.
Zitron is not selling the usual reskill-harder fairy tale. He does a useful autopsy of the money loop: hyperscalers fund AI startups, startups purchase hyperscaler compute, lenders finance the data centers, and Wall Street counts the resulting circulation as organic growth. That part is forensic rather than soothing. But he is still looking at the wreckage through the balance-sheet end of the telescope. The missing noun is worker. In his account, AI is mainly a fraudulent capex trade that will eventually be exposed; the harder question—what happens when the tools can perform paid tasks more cheaply even if their vendors are unprofitable—never arrives. He substitutes financial collapse for social analysis and treats weak economics as evidence against utility, when subsidised systems can still destroy labour demand. Honest about the bubble, incomplete about the obsolescence.