Sumeet Singh argues that many AI founders are mistakenly applying old SaaS models to new AI opportunities. He highlights two viable paths: building infrastructure for AI models or creating workflows unique to AI's capabilities. Emphasizing Richard Sutton's "bitter lesson," he warns that specialization will likely lead to irrelevance.

The article explores the concept of economic bubbles, particularly in tech, arguing that while they can lead to downturns, they also drive significant innovation and infrastructure development. Drawing on historical examples, it highlights how speculative investments can create the conditions for transformative advancements.