Inside the economics of OpenAI (exclusive research)

The podcast explores the economics of AI, questioning whether the high costs of training and running frontier models can translate into a profitable business, drawing parallels to companies like Uber. The discussion highlights Epoch AI's research, which suggests that while OpenAI may have decent gross margins, the rapid obsolescence of models and high R&D spending create challenges. The conversation touches on the potential of advertising revenue for OpenAI, compute costs, and the strategic focus of companies like Anthropic. The panel also discusses the constraints on scaling AI infrastructure, particularly the availability of GPUs, and debates whether energy or chip supply is the greater bottleneck. The potential shift towards running AI models on edge devices and the increasing demand for AI-driven solutions in enterprises are also addressed.