Premium: AI Isn’t Too Big To Fail

Source: Where’s Your Ed At Date: 2026-04-03 URL: https://www.wheresyoured.at/premium-ai-isnt-too-big-to-fail/

Summary

AI is not systemically important enough to be “too big to fail.” Zitron compares AI economics to subprime mortgages — VC-subsidized, not profitable, not load-bearing for the broader economy — and argues that unlike 2008, a collapse would look like the dot-com bust: painful but not catastrophic. He goes further: the failure is “necessary for us to move forward as a society.”

Implications

Removing the implicit backstop argument changes the institutional investor risk calculus. If AI can fail without taking the economy with it, the incentive to hold through a downturn weakens.

• AI financial sustainability. Dot-com framing is useful here — the bust wiped out companies but the internet survived. The question is which AI infrastructure (inference APIs, foundation models, tooling) survives the shakeout vs. what is purely speculative.
• Generative-AI ROI doubt. “Necessary failure” implies Zitron thinks current AI is actively harmful, not just unproductive — the social cost of the hype cycle is part of the argument, not just the financial cost.
• Watch: institutional investor exposure to AI, capital rotation away from pure-play AI companies, and whether “too big to fail” rhetoric starts appearing in vendor communications as a defensive move.