Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Presented by Nvidia
The AI revolution is accelerated by billion-parameter-based models needed to develop agent and physical AI. As the NVIDIA founder and CEO shared the full production of fully production, the transition to the full production of the exercise, as it becomes an AI factory designed to effectively and effectively overcome the world, turns AI to AI factories in AI. To get acquainted with a $ 1 trillion opportunity, NVIDIA in GTC – a operating system for Blackwell Ultra AI platform and AI factories, robotics and accelerated calculation.
Blackwell is already in full production – a surprisingly 40x performance increase over Hopper. This architecture redefines the training and results of AI model, and the AI apps can be more efficient and more expandable. And the second half of the second half, the next evolution of the Blackwell AI factory platform: Blackwell Ultra – GPU GPU to support the next generation of AI models.
Nvidia continues to move fast to the annual AI architecture. Nvidia Vera Rubin AI Information Center performance and efficiency was designed for Supercare.
Outside GPU, AI infrastructure, photonsicks, a seismic turns with innovations in the network and developed network. These improvements will increase expansion, efficiency and energy consumption between mass AI data centers.
Meanwhile, according to Huang, the physical AI for robotics and industry is the $ 50 trillion dollar. The AI-powered automation from production and logistics is preparing to change the entire industry. NVIDIA ISAAC and COSMOS platforms are in the front row by driving the next period of AI-drive robotics.
Some Nvidia ads in GTC
Nvidia Roadmap: NVIDIA Roadmap includes Vera rubin, Toobe released in the second half of 2026, in 2027, Vera Rubin Ultra is bragged with data transfers between the plugs, which are a critical feature for large AI systems, especially with large AI systems. Feynman planned for 2028The next architecture next architecture will be released using the next gene HBM memory.
DGX Private AI Computers: The NVIDIA Grace is designed to develop extensive models, delicate and inferences in the DGX Spark and DGX station design equipped with the Blackwell Platform. They will be produced by a number of companies, including Asus, Dell and HP.
Spectrum-x and quantum-x network Platforms: These Silicone Potonic Network Switches make millions of AI factories to combine GPUs between sites and sharply power consumption. Quantum-X Photonics Infiniband switches This year and spectrum photons will come in 2026 Ethernet keys.
Dynamo Program: The free, open source dynamo program helps to speed up a multi-step justification process, increase efficiency and reduce the update time in AI factories.
Nvidia Accelerated Quantum Research Center: A Boston-based research center will present advanced technologies to develop quantum calculations along with leading apparatus and software manufacturers.
Nvidia isaac gr00t n1: A model created for humanoid robots, GR00T N1 is a fully customizable foundation model of the world’s first open, full-time world for generalized humanoids and skills. There is a binary system similar to justifiable models for both fast and slow thinking.
Newton Physics Engine: Nvidia, Google Deepmind and Disney research, also declared a collaboration with Disney studies to develop an open source physics engine, learning how to manage complex tasks with higher accuracy.
These are just points – Don’t miss the full GTC recap, live on the blog of NVIDIA.
Sponsored articles are generated or clearly clearly produced by a company that has a business relationship with the post of payment or venturebeat. Contact for more information
