Nvidia GTC 2026 Live Blog: Jensen Huang’s Keynote, Hardware Reveals, and More AI News
Jensen Huang unveils Vera Rubin architecture and rack-scale systems for gigawatt data centers.
Nvidia's GTC 2026 conference, led by CEO Jensen Huang, signals a pivotal industry shift from proving AI's potential to building the infrastructure for its industrial-scale execution. The focus is on creating the modern 'AI factory,' moving beyond individual chips to fully integrated systems designed to drastically reduce the cost per token of AI inference. The keynote is expected to provide a full technical breakdown of the next-generation Vera Rubin architecture, which includes the Rubin GPU and a custom ARM-based Vera CPU, highlighting advancements like HBM4 memory for high-performance inference.
Central to this vision are massive, liquid-cooled rack-scale systems engineered to meet the intense power and networking demands of gigawatt-scale data centers. Alongside this hardware push, Nvidia is detailing software frameworks to help enterprises transition from basic chatbots to autonomous agentic AI—systems that can reason and execute complex workflows. The event will also feature major updates to platforms like Nvidia Isaac, merging foundation models with physical robotics and digital twin technology to enable real-world execution in humanoid and industrial settings.
- Vera Rubin architecture debut with Rubin GPU, Vera CPU, and HBM4 memory for next-gen inference
- Rack-scale, liquid-cooled systems unveiled to power gigawatt-scale AI data center demands
- New software frameworks for building agentic AI and updates to Nvidia Isaac for physical robotics
Why It Matters
This defines the hardware and software stack needed to deploy AI at industrial scale, moving from prototypes to production.