Nvidia Corp., the leading chipmaker, recently revealed its upcoming artificial intelligence (AI) superchip, Blackwell graphics processing unit (GPU), in response to the worldwide surge in demand for AI chips. The latest chip reportedly performs some tasks 30 times faster than previous iterations. There was a marginal decline in Nvidia shares during the pre-market activity on Nasdaq, coming in at $858.20.
The announcement came during the GTC conference in San Jose, California, where the company's founder and CEO, Jensen Huang, introduced the NVIDIA DGX SuperPOD. This advanced AI supercomputer uses the NVIDIA GB200 Grace Blackwell Superchips which are designed to process trillion-parameter models continuously for superscale generative AI training and inference workloads.
Huang presented NVIDIA Blackwell as a substantial update to existing AI infrastructure. He stated that the Blackwell platform would catalyze real-time generative AI in trillion-parameter large language models. Blackwell, named after the mathematician David Harold Blackwell from the University of California, Berkeley, will succeed the NVIDIA Hopper architecture launched two years prior.
The NVIDIA GB200 Grace Blackwell Superchip combines two Blackwell NVIDIA B200 Tensor Core GPUs with the NVIDIA Grace CPU, integrating them through a 900GB/s ultra-low-power NVLink chip-to-chip connection. Tech giants like Alphabet, Google, Amazon, Dell, Meta, Microsoft, OpenAI, Oracle, and Tesla have already backed Blackwell. Global cloud service providers, AI pioneers, system and server vendors, and regional cloud service providers from around the world have shown interest in its adoption.
Huang emphasized the need for a new method of computing to keep up with scale, drive down costs, and maintain sustainability. He suggested that accelerated computing could provide the answer.
The CEO also introduced several new software tools, including NIM microservices and Omniverse Cloud APIs. NVIDIA NIM offers developers a platform for implementing custom AI, and Omniverse Cloud APIs provide advanced simulation capabilities.
Huang envisions future data centers as "AI factories," generating intelligence to increase revenues. In the telecommunications sector, the company announced the NVIDIA 6G Research Cloud, an AI and Omniverse-powered platform to progress the next communications era.
This announcement also tackled breakthroughs in semiconductor design and manufacturing with a new computational lithography platform, cuLitho, in conjunction with TSMC and Synopsys. NVIDIA's Earth Climate Digital Twin is a cloud platform that provides interactive, high-resolution simulation for speeding up climate and weather prediction.
Moreover, NVIDIA launched over two dozen new microservices, allowing healthcare enterprises globally to leverage the latest advances in generative AI. Finally, with the introduction of Omniverse to Apple Vision Pro, developers can now stream interactive industrial digital twins into VR headsets. In the robotics sector, BYD, the autonomous vehicle company, will use the next-gen NVIDIA computer for its AV, paving the way for its future EV fleets.
