Market bubbles are not simply the result of founder optimism. They are systemic "consensual hallucinations" where business models incentivize all participants—including underwriters and analysts—to perpetuate the hype, even when fundamentals are weak, as shown by banks issuing 'buy' recommendations during the dot-com peak.
Unlike search or social media, where value was captured by shareholders, AI may be a foundational technology whose benefits primarily "leak" to customers. Similar to jet travel or vaccines, the companies creating the tech may not be the biggest financial winners; the end-users who gain productivity and new capabilities will be.
After initial unrestricted spending led to budget overruns at companies like Uber, major enterprises are shifting focus. They are moving away from measuring raw AI usage (tokens) and toward implementing AI only for proven use cases with clear ROI, which may benefit cheaper, open-source models over expensive frontier ones.
The dot-com bubble burst in a specific sequence: consumer-facing companies failed first, followed by their business-to-business suppliers, and finally the core infrastructure providers. A similar pattern of contagion is predicted for the AI sector, with cracks first appearing in consumer-focused applications.