Source: Financial Times (paywall)
The venture capital market has decoupled from traditional commercialization metrics, flooding AI infrastructure and research plays with capital that dwarfs what foundational model companies raised pre-deployment. Two dynamics collide here: genuine supply-side belief that compute and talent are the only scarce resources that matter, and speculative FOMO that treats AI as a category bet rather than a business bet. The result is unsustainable valuations for entities with no clear path to revenue capture. The gap between this capital velocity and actual product-market fit in AI applications will expose which labs have defensible moats and which are simply well-funded research projects.