AI World at GTC: Open Source, agents and the token economy

For the second consecutive time, Jensen Huang featured AI World's analysis in his GTC keynote. Following GTC Washington D.C. in October 2025, our visualisation, tracking Big Tech's growing contribution to the open-source ecosystem, was on the main stage in San José in front of over 30.000 people. The graph Jensen highlighted was first published in our story Big Tech Is All-In On Open-Source AI, and the data behind it continues to tell a compelling story.

NVIDIA now leads all companies on Hugging Face by number of open-source repositories, a remarkable shift for a company whose identity was built around proprietary hardware. This is not a coincidence, it reflects a strategic positioning that was on full display throughout GTC 2026.

Three themes dominated the conference and are directly relevant to the work we do at AI World.

1) Open source is the foundation. The open models panel, moderated by Jensen Huang and bringing together the CEOs of Mistral AI, Thinking Machines Lab, Perplexity, Cursor, LangChain, Black Forest Labs and others, made one thing clear, open models are no longer an alternative approach. They are increasingly the foundation on which AI is being built. As Mistral AI's Arthur Mensch put it, open models are what allows AI to reach everyone in society, not only those with access to proprietary systems.

2) Agents are the next wave. Every session pointed in the same direction, every company, every individual will use AI agents in the near future. NVIDIA announced NemoClaw, an enterprise-grade distribution of OpenClaw with a built-in security and privacy layer, alongside the AI-Q open agent blueprint and Nemotron 3 Super. The infrastructure for the agent era is being built now.

3) Tokens are the new unit of value. Jensen Huang's clearest message was that every company will soon think about its business in terms of token production, not what software it runs or what data it stores, but what intelligence it generates. This has direct implications for how AI capability will be measured, traded and mapped, which is precisely what AI World is built to track.

GTC 2026 confirmed what the data has been showing for some time, open ecosystems are winning, agents are arriving, and the global distribution of AI capability is shifting fast.