NVIDIA CEO Jensen Huang has unveiled a wide-ranging set of collaborations with leading technology companies, including Google, Microsoft, and Oracle.The CEO of NVIDIA, Jensen Huang, has unveiled a wide-ranging set of collaborations with leading technology companies, including Google, Microsoft, and Oracle, to offer new capabilities in artificial intelligence (AI), simulation, and collaboration across all industries. The companies will offer a complete set of systems and services for startups and enterprises racing to put these innovations to work on a global scale.
At the GTC 2023 event in San Jose, CA, Mr. Huang in his keynote highlighted how industries are confronting powerful dynamics such as digitalization, sustainability, and generative AI, and acceleration is the most effective method for reclaiming power and achieving sustainability and Net Zero. The conference features 650 talks from industry leaders, including Demis Hassabis of DeepMind and Valeri Taylor of Argonne Labs, among others.
NVIDIA’s advancements in the field of AI have been significant, according to Huang, and the company’s DGX supercomputers, which were initially utilized for AI research, are now operating at businesses worldwide to improve data and process AI. Mr. Huang also introduced a new GPU for large-language-model inference, the H100 NVL with dual-GPU NVLink, built on NVIDIA’s Hopper architecture, which can help cut the expenses of processing massive language models by an order of magnitude.
NVIDIA AI Enterprise is an acceleration software package for end-to-end development and implementation of AI. NVIDIA DGX Cloud is optimized to operate NVIDIA AI Enterprise, and partnerships with industry-leading cloud service providers, such as Oracle Cloud Infrastructure and Microsoft Azure, will house the DGX Cloud infrastructure.
NVIDIA AI Foundations is a collection of cloud services that enables customers to create, improve, and administer bespoke LLMs and generative AI systems that have been trained with their confidential data and for domain-specific tasks. In addition, NVIDIA introduced inference platforms for AI video and picture generation, as well as LLM implementation and recommender inference, with the goal of assisting businesses in their deployment of quickly developing generative AI models.
NVIDIA Omniverse Cloud
NVIDIA has also introduced two new cloud services, NVIDIA Omniverse Cloud and NVIDIA Grace Hopper for Recommendation Models (L4), which are available through Google online. L4 excels at graph recommendation models, vector libraries, and graph neural networks.
NVIDIA cuLitho, a technological breakthrough that introduces accelerated processing to the field of computational lithography, is currently being incorporated by TSMC, TSMC, and Synopsys into their software, manufacturing processes, and systems for the most recent generation of NVIDIA Hopper architecture GPUs.
In addition, AT&T is evaluating Omniverse ACE and the Tokkio AI character pipeline to create, personalize, and implement virtual companions for customer support and its employee help center. NVIDIA AI is also being used by businesses worldwide to process data more effectively, with 40,000 of NVIDIA’s customers utilizing the high-performance NVIDIA TensorRT inference scheduler and runtime, as well as the NVIDIA Triton multi-framework data center inference providing software.
NVIDIA has released new versions of one hundred acceleration libraries, including cuQuantum, designed for quantum computing; cuOpt, designed for combinatorial optimization; and cuLitho, designed for computational lithography. There are currently 4 million developers, 40,000 businesses, and 14,000 entrepreneurs that make up the NVIDIA community worldwide.
Huang expressed gratitude to NVIDIA’s collaborators in the systems, cloud, and software industries, as well as the company’s researchers, scientists, and employees, for enabling the world to achieve what was once thought to be unachievable.