NVIDIA Computex 2026 Keynote: Agentic AI, Vera Rubin Infrastructure, PC Reinvention
Generated: 2026-06-02 · API: Gemini 2.5 Flash · Modes: Summary
NVIDIA Computex 2026 Keynote: Agentic AI, Vera Rubin Infrastructure, PC Reinvention
Clip title: Nvidia’s Computex 2026 Keynote in Less Than 12 Minutes Author / channel: CNET URL: https://www.youtube.com/watch?v=ugNnw4lAMWA
Summary
Jensen Huang, CEO of NVIDIA, opened his GTC Taiwan keynote by declaring the arrival of “agentic AI” – intelligent software agents capable of performing complex, productive tasks. He detailed this new computing pattern where users explain their intent to an AI agent, which then generates code or utilizes tools to achieve the desired outcome, replacing traditional click-and-type interactions. This represents a fundamental shift towards a disaggregated and distributed computing model, where AI agents, powered by large language models (LLMs) and managed by a “harness,” will orchestrate work across various domains, revolutionizing how we interact with technology and solve problems.
NVIDIA is positioning itself as a “full-stack AI infrastructure company,” moving beyond just GPUs to provide comprehensive systems. Huang introduced “Vera Rubin,” describing it not as a single chip but an entire end-to-end AI infrastructure. This includes Vera Rubin GPUs, NVLink 72, Vera CPUs, Spectrum-X switches, Grace CPU trays, and BlueField-4 storage trays, all designed to build “AI factories.” Notably, the “Vera” CPU is purpose-built for AI agents, contrasting with past CPUs designed for humans. It features incredible bandwidth (3.6 terabytes per second) and is integrated with Vera Compute for orchestration and Vera BlueField Storage for high-speed data access, showcasing NVIDIA’s ambition to provide complete, liquid-cooled AI supercomputers.
A significant announcement was NVIDIA’s collaboration with Microsoft to “reinvent the PC.” This new era of personal computing will be powered by “NVIDIA RTX Spark,” a chip integrating a powerful Blackwell RTX GPU (with 6,144 CUDA Cores and 1 PetaFLOP of AI performance) and a custom 20-core Grace CPU developed with MediaTek, all fused with 128GB of unified memory via NVLink. This integration will enable Windows PCs to run both traditional applications and sophisticated AI agents, offering unprecedented performance and memory (with desktop models boasting up to 768GB RAM and 20 PetaFLOPS of AI compute). Huang envisioned PCs evolving into intelligent companions, akin to R2D2 or C3PO, capable of running trillion-parameter models right on your desk.
Finally, NVIDIA is extending its AI agent vision to robotics and autonomous systems, ushering in the “age of physical AI.” The company unveiled “Cosmos 3,” a foundation model for physical AI that enables robots to observe the physical world, understand events, and generate physics-accurate synthetic videos for training. For autonomous vehicles, “Alpamayo 2” was introduced as an open model for self-driving cars, with NVIDIA’s DRIVE Hyperion platform already adopted by over 80% of global automakers. Furthermore, NVIDIA announced the “Isaac GROOT Humanoid Robot Platform,” an open development platform for humanoid robots, featuring Isaac Simulators, GROOT Open Models, and tools for training data generation, all running on the Isaac OS, providing a complete ecosystem for developing sophisticated robots.
Video Description & Links
Description
Nvidia CEO Jensen Huang took the stage at GTC to unveil the new Vera Rubin AI computing platform, the Vera CPU, a new collaboration with Microsoft that promises to “reinvent” the PC and its new open source AI models.
Add CNET as a trusted news source https://www.google.com/preferences/source?q=cnet.com Never miss a deal again! See CNET’s browser extension 👉 https://bit.ly/3lO7sOU Check out CNET’s Amazon Storefront: https://www.amazon.com/shop/cnet Subscribe to CNET on YouTube: https://www.youtube.com/cnet Follow us on TikTok: https://www.tiktok.com/@cnetdotcom Follow us on Instagram: https://www.instagram.com/cnet Follow us on Bluesky: https://bsky.app/profile/cnet.com Like us on Facebook: https://www.facebook.com/cnet CNET’s AI Atlas: https://www.cnet.com/ai-atlas Follow us on X: https://www.x.com/cnet Visit CNET.com: https://www.cnet.com
Tags
CNET, Technology, tech, Tech Devices, nvidia keynote, nvidia keynote supercut, jensen huang, nvidia, nvidia rubin, nvidia vera rubin, jensen huang keynote, nvidia blackwell, nvda, nvidia ces, nvidia tech news, nvidia ai, nvidia graphics card, generative ai, artificial intelligence, nvidia robotics, humanoid robots, nvidia stock, tech keynote highlights, nvidia summary, nvlink, nvidia architecture, future of ai, computing news, Computex 2026, Supercut
URLs
- https://www.google.com/preferences/source?q=cnet.com
- https://bit.ly/3lO7sOU
- https://www.amazon.com/shop/cnet
- https://www.youtube.com/cnet
- https://www.tiktok.com/@cnetdotcom
- https://www.instagram.com/cnet
- https://bsky.app/profile/cnet.com
- https://www.facebook.com/cnet
- https://www.cnet.com/ai-atlas
- https://www.x.com/cnet
- https://www.cnet.com