In an interview with Bloomberg Television during a visit to South Korea, Nvidia Corp CEO Jensen Huang argued that proprietary AI models can be cheaper than open models, challenging the common belief that open systems are a lower-cost alternative. Huang stated, “Frankly, I think closed models are cheaper,” explaining that companies often overlook hidden costs of deploying open models, including the expertise needed to fine-tune, maintain, guardrail, keep safe, evaluate, and build computers to host them. He said the total cost of ownership for self-hosting can exceed paying for a hosted proprietary model. Huang clarified that open models are necessary for control, adaptation to specialized use cases, protecting proprietary knowledge, meeting regulatory requirements, or maintaining technological sovereignty. Otherwise, he suggested using commercial models through the cloud is often simpler, noting he encourages businesses—including Nvidia—to use services like OpenAI, Claude, Cursor, Cognition, and Perplexity whenever practical because “it’s just easier,” adding that companies should “build only what you must.”