NVIDIA announcements, news and more, from GTC 2025

0 3 minutes read

Nafidia presented it

The artificial intelligence revolution is accelerating, driven by the teaching of the teaching of thinking necessary to develop artificial intelligence agents and materialism. Since the founder of NVIDIA and CEO of JenSen Huang participated in his GTC keynote, the transition from training to inference is complete production causes demand at the expense of artificial intelligence to rise with the transformation of data centers around the world to Amnesty International factories designed to clarify millions of user inquiries efficiently and effectively. To meet this 1 trillion dollar opportunity, NVIDIA in GTC has revealed great progress – from the Blackwell Ultra AI platform and an operating system for artificial intelligence factories to progress in communication, robots and accelerating computing.

Blackwell is already in full production – an amazing batch of 40x performance on Hopper. This structure redefines and inferred training on the artificial intelligence model, which makes artificial intelligence applications more efficient and more expandable. The coming in the second half 2025 is the following development of the Blackwell Ai Factory: Blackwell Ultra- Strong graphics processing unit with extensive memory to support the next generation of artificial intelligence models.

NVIDIA continues to move quickly, committed to updating annual architecture. Nvidia Vera Rubin is designed to charge performance and efficiency at the AI Data Center.

Beyond graphics processing units, Amnesty International’s infrastructure is subject to a seismic transformation with optical innovations, improved storage and advanced networks. These breakthroughs will significantly enhance expansion, efficiency and energy consumption through huge artificial intelligence centers.

Meanwhile, material artificial intelligence of robots and industry is an enormous opportunity of $ 50 trillion, according to Hanging. From manufacturing and logistical services to health care and beyond, automation that works from artificial intelligence is preparing to reshape the entire industries. NVIDIA ISAAC and Cosmos platforms are in the foreground, leading the next era of robots driven by artificial intelligence.

Some NVIDIA ads in GTC

NVIDIA Road Map: Includes NVIDIA Road Map Vera Robin, TOBE was released in the second half of 2026, followed by Vera Rubin Ultra in 2027. Rubin and servers are characterized by improved speeds, especially in databases between chips, an important feature of large AI systems that contain many chips. It is scheduled for 2028 is FeynmanThe next structure that will be released, taking advantage of the HBM memory from the next generation.

AI Personal Computers: DGX: Supported by the Nvidia Grace Blackwell platform, DGX Spark and DGX station designed to develop large models, adjust and infer them on desktop computers. It will be manufactured by a number of companies, including ASUS, Dell and HP.

Spectrum-X and Quantum-X Networks Platforms: Silicon optical networks help these AI factories connect millions of site graphics processing units and significantly reduce energy consumption. Infiniband quantum keys will be available later this year, and the ETHERNET optical keys will reach Spectrum-X in 2026.

Dynamo Program: The free and open source Dynamo program helps to accelerate the multi -step thinking process, improve efficiency and reduce time to innovate in artificial intelligence factories.

Right quantum research center in Nafidia: The Boston -based research center will provide advanced technologies to enhance quantum computing in cooperation with leading hardware and programs.

Nvidia Isaac Gr00T N1A basic model for human robots, GR00T N1 is the first primary and fully customized basis model for general thinking and human skills. It contains a dual system similar to thinking forms, for both fast and slow thinking.

Newton’s physics engine: NVIDIA has also announced cooperation with Google DeepMind and Disney Research to develop Newton, an open source physics engine that allows robots to learn how to deal with complex tasks more accurately.

This is just the most prominent – Don’t miss the full GTC extract, live on the NVIDIA blog.

Articles sponsored by content produced by a company that pays the price of the post or has a commercial relationship with Venturebeat, which are always clearly specific. For more information, call sales@venturebeat.com.