20VC: Is Anthropic's Coding Business Worth $2 Trillion? | Should American Enterprises Work With Open-Source Chinese Models? | Why 80–90% of Neo-Labs Die in the Next 18 Months? with Eno Reyes, Co-Founder @ Factory

The future of AI infrastructure centers on the transition from general-purpose frontier models to specialized, open-source systems that prioritize outcome-based efficiency over raw token costs. Enterprises must secure "sovereign intelligence" by maintaining control over their proprietary workflows and internal harnesses, rather than relying solely on external model providers that may eventually become direct competitors. Most current AI startups face an existential threat, with 80-90% likely to fail within 18 months as they lack durable, defensible moats. True value in the coming years will emerge not from replacing human labor, but from creating entirely new software development methodologies where AI acts as a foundational component within a company's own systems. Ultimately, the most successful businesses will be those that treat AI as a tool for resource allocation and outcome optimization rather than a black-box service.