• •SemiAnalysis reports that value capture in the AI stack is increasingly consolidating at the model layer rather than the application layer.
• •The analysis highlights that while application-layer startups face high customer acquisition costs and low switching barriers, foundation model providers benefit from platform-level network effects.
• •It remains unclear whether enterprise-specific models or agentic workflows will eventually provide enough differentiation to allow application-layer firms to retain long-term margins.

SemiAnalysis reports that value capture in the artificial intelligence sector is shifting from thin application wrappers to foundational model providers. Unlike previous software cycles where data moats fueled application dominance, foundation models currently benefit from superior scale and infrastructure integration. However, the report notes that application-layer startups struggle with high customer churn and diminishing differentiation as AI capabilities become commoditized. Whether application providers can eventually secure a sustainable competitive advantage will depend on their ability to integrate proprietary enterprise workflows that models cannot easily replicate.