Nvidia, the world’s third most valuable company by market cap, kicked off its GPU Tech Conference (GTU) on March 18th, in San Jose, California.
The annual conference, which runs through Thursday, is an eagerly anticipated event by investors and developers who are keen to learn about the chipmaker’s artificial intelligence initiatives and partnerships.
On Monday, CEO Jensen Huang started his keynote speech by reminding attendees of Nvidia’s original mission, which is “computer graphics, physics, and artificial intelligence, all intersecting inside of a computer”.
He made some important announcements during the first day of the GTU, which we will be discussing here.
The Blackwell Superchips
Nvidia unveiled the Blackwell platform, the latest version of its AI-powered graphics processing units (GPU). Huang said Blackwell is the world’s most powerful chip that will power “a new era of computing” by “pushing the limits of physics of how big a chip could be”.
The company says the system is more capable and efficient than its predecessor, Hopper.
Huang claims that Blackwell is 30 times faster than Hopper and is capable of running a full-scale generative AI large language model (LLM) while consuming around 25 times less cost and energy than its predecessor.
Nvidia’s president explained that it took 8,000 GPUs, 15 megawatts, and 90 days to create the GPT-MoE-1.8T model.
The chipmaker also unveiled the GB200 Grace Blackwell Superchips, which it says is a “massive” chip that connects two NVIDIA B200 Tensor Core GPUs to the NVIDIA Grace CPU.
The super chip will also be available on NVIDIA DGX Cloud, an AI-powered platform co-engineered with leading cloud service providers that gives enterprise developers access to the infrastructure and software required to build and deploy advanced generative AI models.
Several leading AI companies are already set to use the Blackwell system ahead of its official launch, including Amazon, Microsoft, and Google.
Nvidia Omniverse is Coming to VisionPro
Nvidia announced that it is bringing Omniverse to Apple’s recently launched mixed-reality headset, the VisionPro. Omniverse is the company’s computing platform that enables individuals and teams to develop 3D workflows and applications.
During the keynote, Huang demonstrated Omniverse simulations where developers can send their Universal Scene Description (OpenUSD) industrial scenes from content creation applications to the NVIDIA Graphics Delivery Network (GDN), which will then stream “advanced 3D experiences” to the Apple Vision Pro.
GDN is a global network of graphics-ready data centers that can project 3D experiences to virtual reality headsets.
AI Humanoid Robots
Nvidia then showcased Project GR00T, a foundational model for humanoid robots that Huang said was the “next wave” of AI and robotics systems.
The Generalist Robot 00 Technology or GR00T is designed from the ground up to understand natural language and emulate human actions and movements from observation.
The chipmaker also unveiled ‘Jetson Thor’, an all-new computer for humanoid robots that has generative AI capabilities.
Next-Gen Data Centers
Nvidia showed a 3D blueprint for the next generation of data centers in the AI era, the design of which was created using Omniverse.
Huang noted that in the future, data centers will be called AI factories and the goal of such a facility will be the “generation of intelligence”.
6G Cloud Platform
The company announced a 6G research platform that is said to offer researchers a “comprehensive suite” to advance AI for radio access network (RAN) technology.
In a press release, Nvidia said the platform allows organizations to “accelerate” the development of 6G technologies that will connect trillions of devices with cloud infrastructures.
According to the chipmaker, this will lay the foundation for a “hyper-intelligent world” supported by autonomous vehicles, smart spaces, and a wide range of extended reality and immersive education experiences and collaborative robots.
Nvidia claims its 6G cloud platform will empower the next generation of wireless technology by meeting the growing demand for running AI on wireless devices.
Nvidia and Oracle Partner for AI Sovereignty
Nvidia is expanding its partnership with Oracle to help deliver sovereign AI globally. Sovereign AI refers to a nation’s capability to produce artificial intelligence using its own infrastructure, data, workforce, and business networks.
The partnership is meant to deliver sovereign AI solutions to customers around the world through Oracle’s distributed cloud, AI infrastructure, and generative AI services, and Nvidia’s accelerated computing and generative AI software.
The goal of the program is to enable governments and enterprises to set up AI factories or data centers across the world.
