Ed Zitron Skewers AI Bubble Economics

Strong thesis, bluntly stated: the AI stack looks less like a durable platform shift and more like a subsidy chain that breaks as soon as anyone tries to charge true cost.

• –Cursor is the clearest example of the pressure point: a high-growth AI coding business that depends on upstream model pricing it does not control.
• –The article’s core risk is vendor lock-in disguised as convenience; once rate limits tighten, product value drops and churn rises.
• –Local and self-hosted OSS models matter here because they reduce exposure to API price hikes, throttling, and feature degradation.
• –The broader warning is that many AI subscriptions only work while capital is masking usage economics, which is not a stable long-term market structure.
• –For developers, the practical takeaway is to diversify model providers and keep an offline/self-hosted fallback in the loop.