NVIDIA’s annual GTC conference, often called the “Woodstock of AI,” kicks off in San Jose this week. Past editions celebrated the raw power of generative AI, but this year Jensen Huang is expected to shift the conversation toward Agentic and Physical AI. NVIDIA is no longer just a chipmaker. It wants to be the architect of the world’s AI infrastructure. Here is what to look forward to.
Survey
The Vera Rubin platform
The Blackwell successor finally gets its full debut. Vera Rubin is named after the pioneering astronomer and is built as a six-chip ecosystem rather than a single GPU. The Rubin GPU uses HBM4 memory and is expected to cut inference token costs by up to 10x. The Vera CPU pairs it with 88 custom cores. With the architecture already in production, Huang is expected to give firm shipping timelines to hyperscale partners like Microsoft and Meta.
NemoClaw and AI agents
Software is NVIDIA’s second act. NemoClaw is an open-source platform built for Agentic AI, where models do not just generate text but plan, reason, and execute tasks across complex workflows. NVIDIA wants its software stack to function as the operating system for enterprise AI agents, reducing the need for constant human input.
The N1X Superchip for AI PCs
Also read: Iran-Israel-US war: AI images and videos intensifying fog of war in Middle East
This is the mystery chip reveal people are watching. The N1X is an Arm-based SoC built for high-end laptops and desktops, developed with MediaTek. It is NVIDIA’s push into the consumer CPU market and a direct challenge to Apple’s M-series and Qualcomm’s Snapdragon X Elite. The goal is to bring data-center-level AI performance to a local workstation.
Project GR00T and Physical AI
Physical AI, meaning AI that operates in the real world, is a central theme this year. Expect updates to Project GR00T, NVIDIA’s foundation model for humanoid robots, alongside Omniverse Digital Twins. Sessions will cover deep reinforcement learning and the mechanics of giving robots human-like balance and spatial awareness.
Gigawatt-Scale AI Factories
Huang has been using the term “AI Factory” to describe the next generation of data centers. GTC 2026 will lay out blueprints for facilities built to handle the extreme power and cooling demands of the Rubin architecture. NVIDIA is working with partners like Vertiv to turn these into repeatable, system-engineered products, treating AI infrastructure less like a bespoke project and more like industrial manufacturing.
Also read: This guy saved his dog from cancer by creating a mRNA vaccine using ChatGPT
Follow Us
Vyom Ramani
A journalist with a soft spot for tech, games, and things that go beep. While waiting for a delayed metro or rebooting his brain, you’ll find him solving Rubik’s Cubes, bingeing F1, or hunting for the next great snack. View Full Profile
