“When new markets have to be created, they have to be created starting here, at the centre of the computer ecosystem,” said Huang (pictured)..
He also announced volume production of an upgraded
RTX Pro Server system with 4x better performance than H100 running DeepSeek and 1.7x better running Llama models.
A new version of its NV Link interconnect technology called NVLink Fusion lets customers use other companies’ CPUs alongside Nvidia GPUs or use Nvidia CPUs with another company’s accelerators..
The successor to its Rubin architecture, which is due next year, will be called Feynman and will ship in 2028.
Nvidia’s workstation for researchers called DGX Spark running the GB10 Grace Blackwell processor delivering 1000 AI TOPS of AI performance will be in production in “a few weeks”, said Huang.
Huang also announced an update to its foundation model for humanoid reasoning and skills Isaac GR00T-Dreams. “Physical AI and robotics will bring about the next industrial revolution,” said Huang, “from AI brains for robots to simulated worlds to practice in or AI supercomputers for training foundation models, NVIDIA provides building blocks for every stage of the robotics development journey.”
