Is the generative AI bubble finally bursting

gary marcus has coined the term 'generative ai fizzle' to describe a potential slow decline in ai valuations, arguing that the market is experiencing a widening gap between hype and actual profits. he cites recent stock market dips, significant losses at openai despite high ad spend, and the commoditization of llms as evidence that the ai sector lacks a defensible moat. marcus highlights that new open-source chinese models and intense price wars are threatening the profit models of us-based giants, making current high-leverage investments increasingly risky. he warns that even if the technology remains, the financial bubble surrounding it—described by paul gambles as the biggest in history—may deflate rather than burst.