Key insights

• ⚙️ Introduction of Blackwell platform with Hopper chip, featuring 208 billion transistors and two dies acting as one giant chip, and 10 terabytes per second data transfer
• 🔌 Form-fit-function compatibility with Hopper for efficient ramping, showcasing significant computational power in a small space
• 🔄 Introduction of MVY Link switch with 50 billion transistors, capable of creating a system with multiple GPUs communicating at full speed
• 🤝 Partnerships with AWS, Google, and other companies for accelerated computing in AI applications, including content token generation using FP4 format
• 📦 Optimization and packaging of pre-trained models for large-scale GPU deployment, focusing on accelerating data processing, AI, databases, and healthcare on Azure
• 🔧 Collaboration with world-leading companies like sap, service now, cohesity, and snowflake to build AI assistants and agents using Nemo and Nims, with optimized software packages for GPUs
• 🏭 Partnership with net app, Dell, and Azure for AI factories and robotics, Omniverse for simulating the digital world and seamless data integration across departments
• 🤖 Showcasing Groot, a humanoid robot learning model powered by Jetson Thor robotics chips, Blackwell and plat processors for AI-powered robotics, and integration with language models

Q&A

• What is Groot, showcased by NVIDIA, and what technologies enable its capabilities?

  Groot is a humanoid robot learning model that can understand and execute instructions. It is powered by technologies such as Isaac Lab, Osmo, and Jetson Thor robotics chips. The model is trained on Omniverse Isa X Sim with Isaac lab and osmo for simulation and training coordination, and it integrates with a language model to generate motions from natural language instructions.

• How is NVIDIA partnering with NetApp, Dell, and Azure to advance AI and robotics?

  NVIDIA is partnering with NetApp, Dell, and Azure to build AI factories, robotics, and Omniverse for simulating the digital world, enabling productivity and seamless data integration across departments. Omniverse is hosted in the Azure Cloud, facilitating the creation of AI-powered robotics and seamless data integration.

• What collaborations are taking place at the Nvidia AI Foundry?

  NVIDIA AI Foundry is collaborating with leading companies like SAP, ServiceNow, Cohesity, and Snowflake to build AI assistants and agents using Nemo and Nims. They offer optimized software packages for GPUs and easy-to-use APIs available for download from Nvidia's website.

• How is NVIDIA optimizing and accelerating cloud computing services through its partnerships with Google, Oracle, and Microsoft?

  NVIDIA is partnering with Google, Oracle, and Microsoft to optimize and accelerate cloud computing services, focusing on accelerating data processing, AI, databases, and healthcare on Azure. Additionally, the company is packaging and optimizing pre-trained models for large-scale GPU deployment as a key goal.

• What are the key features of NVIDIA's new processor for the generative AI era?

  NVIDIA's new processor for the generative AI era utilizes FP4 format for content token generation and introduces the MVY Link switch with 50 billion transistors, enabling the creation of systems with multiple GPUs communicating at full speed. NVIDIA is also partnering with AWS, Google, and other companies to integrate accelerated computing into various AI applications.

• What is the Blackwell platform introduced by NVIDIA?

  The Blackwell platform is a revolutionary platform featuring the powerful Hopper chip with 208 billion transistors and two dies that act as one giant chip, offering 10 terabytes per second data transfer. It is form-fit-function compatible with Hopper, which revolutionizes computational power and size.

• 00:00 NVIDIA introduces the Blackwell platform featuring the powerful Hopper chip with 208 billion transistors, two dies that act as one giant chip, and 10 terabytes per second data transfer. Blackwell is form-fit-function compatible to Hopper, revolutionizing computational power and size.
• 02:14 Nvidia has developed a new processor for generative AI era and a high-performance chip called the MVY Link switch. The company is working with partners like AWS and Google to integrate accelerated computing into various AI applications.
• 04:22 Nvidia and its partnerships with Google, Oracle, and Microsoft to optimize and accelerate various aspects of cloud computing services including data processing, AI, databases, and healthcare on Azure. Nvidia's focus on optimization and packaging pre-trained models for large-scale GPU deployment is a key goal.
• 06:20 Nvidia AI Foundry is collaborating with leading companies like sap, service now, cohesity, and snowflake to build AI assistants and agents using Nemo and Nims.
• 08:22 Nvidia is partnering with net app, Dell, and Azure to build AI factories, robotics, and Omniverse for simulating the digital world, enabling productivity and seamless data integration across departments.
• 11:40 NVIDIA showcases Groot, a humanoid robot learning model that can understand and execute instructions, with the help of technologies such as Isaac lab, osmo, and Jetson Thor robotics chips.