20VC: Satya Nadella on Open vs Closed Models: Who Wins | Whether to Trust Frontier Model Providers | Why Companies Need to Own Their Own Intelligence | The Importance of Verticalisation: From Chips to Models and More

Scaling laws alone do not guarantee AI alignment, necessitating a shift toward engineered trust architectures to contain potentially malfunctioning agents. Organizations must implement robust systems—including identity verification, tamper-proof logging, and behavioral monitoring—to maintain security in an era of increasingly capable models. As AI adoption spreads, firms will differentiate themselves by developing private evaluation benchmarks, effectively turning their unique knowledge into proprietary intellectual property. This transition moves beyond simple model usage toward an implementation economy where businesses integrate AI to optimize complex tasks, from supply chain management to software development. Ultimately, AI serves as a general-purpose technology with the potential to significantly inflect global GDP growth by enhancing productivity and patient outcomes across diverse sectors, provided that companies successfully retain the value created by their internal intelligence rather than leaking it to model providers.