Jensen Huang at GTC: 'Proprietary vs. Open Is Not a Thing — It's Proprietary and Open'

The debate between open-source and closed AI models has dominated industry discourse for the past two years. At GTC 2026, NVIDIA CEO Jensen Huang offered a characteristically direct answer: it's the wrong question.

"Proprietary versus open is not a thing. It's proprietary and open," Huang told attendees at a special GTC session on open frontier models — a framing that captures NVIDIA's increasingly central role in both camps simultaneously.

Why Both Sides Need Each Other

NVIDIA's argument is structural: the diversity of AI models — large and small, open and proprietary, generalist and specialist — is not a market anomaly. It's the necessary architecture for a world where every application will be powered by AI.

Every industry — healthcare, finance, manufacturing — faces unique challenges that require AI systems tuned to their specific data, workflows, and regulatory environments. A single massive frontier model cannot serve all of these needs. The future, according to NVIDIA, is systems of models: specialized components working together, drawing from both open and proprietary foundations.

NVIDIA Is Now the Largest Organization on Hugging Face

To underscore its open-source commitment, NVIDIA disclosed a striking statistic: the company is now the largest organization on Hugging Face, with nearly 4,000 team members contributing to the platform. This is a deliberate strategic position — by contributing models, infrastructure tools, and training frameworks to the open ecosystem, NVIDIA ensures that open-source AI development runs on NVIDIA hardware and uses NVIDIA tooling.

The Nemotron Coalition

The headline announcement from the open AI session was the NVIDIA Nemotron Coalition — a first-of-its-kind global collaboration of model builders and AI labs working to advance open, frontier-level foundation models through shared expertise, data, and compute.

The coalition directly targets the compute barrier that has historically given proprietary labs their advantage: by pooling resources, member organizations can train models at scales that would be unaffordable individually. The Nemotron model family, NVIDIA's open frontier model line, will serve as the shared foundation.

The announcement signals that NVIDIA's competitive strategy extends beyond hardware. By positioning itself as the infrastructure provider for both open and closed AI ecosystems — and now as a coalition organizer for open frontier research — NVIDIA is building lock-in at the platform layer that transcends any individual model cycle.