NVIDIA’s GTC 2026 keynote is today. And the AI industry is tuned in- new chips, new software, and a CEO who knows exactly how to work a crowd.
Jensen Huang is all set to make history on the floor of the SAP Center in San Jose on Monday to deliver his keynote across 30k attendees from 190 countries.
It’s no longer a tech conference but a coronation.
Huang’s presentation covers NVIDIA’s push into AI inference, with new chips and software for autonomous agents. That matters. NVIDIA already commands an estimated 80% of the AI training market share. Inference is the next frontier, and as of now, Google, Amazon, and others are competing rigorously with custom chips. Huang wants that territory too.
He promised “a chip that will surprise the world” and teased “a few new chips the world has never seen before.” Bold word- but they better deliver.
GTC 2026 is where NVIDIA officially kicks off its Vera Rubin platform, replacing Blackwell and Blackwell Ultra. On the software side, NVIDIA is expected to unveil NemoClaw, an open-source platform for enterprise AI agents that offers businesses the right structure to build and deploy AI software.
Then there’s Groq. It’s the first major showcase since NVIDIA’s $20 billion licensing deal with the inference company in late 2025. Everyone wants to know how that integration actually works.
The broader picture is straightforward. NVIDIA is just selling chips, but it’s not merely that. It’s selling the whole stack: hardware, software, models, infrastructure. The company’s announcements today will influence technology roadmaps across the global semiconductor and server supply chains.
No other company in AI has that kind of reach right now. That’s the real story from San Jose.
