Jensen Huang says Open Source can be MORE expensive
Jensen Huang argues that closed models often prove more cost-effective than open-source alternatives when accounting for the hidden expenses of training, fine-tuning, and maintaining safety guardrails. While market narratives suggest open-source tokens have caught up in intelligence and offer lower prices, the total cost of ownership for self-hosted solutions frequently exceeds the price of accessing frontier models. This economic reality explains why revenue continues to concentrate among major labs despite the perception of open-source parity. Consequently, token consumption is rising for open-source providers while the share of economic value flows to proprietary developers.
Dispatches from Iceland
Educators and students in Iceland are navigating a 'tsunami of change' as AI integration reshapes vocational and academic environments, creating a divide between those who view it as 'slop' and those leveraging it for efficiency. While some teachers fear cheating and academic degradation, others have adopted AI as a collaborative tool to generate interactive materials and streamline tasks in fields ranging from automotive repair to hairdressing. The core challenge remains balancing the technology's ability to accelerate work against the risk of eroding human creativity and independent thought, prompting a shift toward using AI as a teaching aid rather than a substitute for learning.