Nvidia CEO Jensen Huang said Blackwell B200 GPUs will be shipped to customers in the fourth quarter of 2024.
Nvidia AI Summit India event was kicked off by company CEO Jensen Huang on Thursday. The CEO took the stage adorned in his trademark black leather jacket and greeted India saying, “India is very, very dear to the world’s computer industry and central to the IT industry.” During his keynote speech, Huang spoke about the rise of technologies such as generative artificial intelligence (AI), deep learning models, and computer vision models, and their global impact. He also walked the crowd through the Nvidia tech stack including the Nvidia Inference Microservices (NIM), and the Blackwell B200 GPU.
Jensen Huang Addresses the Crowd
Huang took the stage to explain the rise of AI. He highlighted that Moore's law, which states that the number of transistors in an IC doubles roughly every two years, is coming to an end. A direct impact of this would be that technology depreciation would stop, and access to technology would be more expensive.
Continuing along the same vein, the Nvidia CEO highlighted that AI could help kickstart a software-based growth journey for technology which can ensure that technology continues to grow and depreciate at the same rate. The Nvidia CEO also highlighted that the company has a strong AI infrastructure in India which is bolstering the AI ecosystem in the country.
- IBM Releases New AI Models for Businesses as GenAI Competition Heats Up
Coming to the company's tech stack, Huang highlighted Nvidia's Compute Unified Device Architecture (CUDA) system. He claimed that the technology has accelerated many important industries such as semiconductors, manufacturing, computation, and more.
Nvidia's Blackwell GPUs Will Be Shipped to Customers in Q4
A major announcement made by Huang is that the Blackwell B200 GPUs will be shipped to customers in the fourth quarter of 2024. Notably, a design flaw was spotted in the chipsets recently which delayed the shipping of the latest Nvidia chips aimed at hardware acceleration of AI workflows. The company took full responsibility for the flaw, and on Wednesday, the company announced that the snag had been fixed.
Highlighting Nvidia's India partners, Huang named Tata Communications, Infosys, Flipkart, Reliance, and others, who are currently leveraging Nvidia's tech stack to build solutions for India and the rest of the world. The company is also investing in building Hindi-based large language models (LLMs) in India with local startups, enterprises, and other stakeholders. “Once you have cracked LLMs in India, you can build in any part of the World,” Huang added.
- US Weighs Capping Exports of AI Chips From Nvidia, AMD to Some Countries
- Nvidia CEO Says ‘Reasoning’ AI Will Depend on Cheaper Computing
The Nvidia CEO also spoke about agentic AI models, which are specialised chatbots capable of handling complex tasks end-to-end. Huang said the company is now investing in developing platforms that let other companies build this capability for themselves. Highlighting a benefit, he said that agentic AI models will help employees become “super employees” and improve productivity.