Nathan Lambert argues that open and closed AI models are on fundamentally different growth trajectories. Closed frontier labs like Anthropic and OpenAI have found strong product-market fit with coding agents, where users willingly pay premium prices for marginally better intelligence. This creates a durable oligopoly at the top end of knowledge work. Meanwhile, open models will grow more slowly but ultimately capture far greater total market value by serving enterprises with commodity pricing, niche tasks, and customizable deployments. The two ecosystems are not in direct competition — they serve different economic niches and will coexist, with closed labs resembling a hybrid of Apple and Microsoft, and open model infrastructure becoming a broad, diverse stack of commodity providers.
