Nvidia CEO Jensen Huang introduced the new Blackwell graphics processing unit (GPU) at the annual GTC conference in San Jose, California, saying it will drive the next era of computing.
During the presentation, he also highlighted the Nvidia Quantum-X800 and Spectrum-X800 communications technologies developed in Israel.
The Globes reported earlier this year that NVIDIA's second-largest development center outside the United States is in Israel, with about 3,300 employees.
Blackwell GPUs, combined with new NVLink technology also developed in Israel, are making waves in AI infrastructure. The company said this will enable organizations to build and run real-time generative AI on large language models with trillions of parameters at up to 25x less cost and energy consumption than previous generations.
The Blackwell GPU architecture features six innovative technologies for accelerated computing that enable breakthroughs in generative AI, data processing, engineering simulation, electronic design automation, and computer-aided drug design. Helpful. These are all new industry opportunities for Nvidia.
“For 30 years, we have pursued accelerated computing with the goal of enabling transformative breakthroughs such as deep learning and AI,” Huang said at GTC. “Generative AI is the defining technology of our time. Blackwell is the driving force behind this new industrial revolution. By working with the world’s most dynamic companies, we are unlocking the potential of AI in every industry. We will make it happen.”
Many organizations expected to adopt Blackwell include Amazon Web Services, Dell Technologies, Google, Meta, Microsoft, OpenAI, Oracle, Tesla, xAI, and more. The company said Blackwell-based products are expected to be available from partners later this year.
What is Blackwell?
Blackwell, which the company claims is the world's most powerful chip, replaces Nvidia's Hopper (H100 and H200) GPUs, which have become the go-to GPUs for AI applications. Blackwell will be available as a standalone GPU, or by combining two Blackwell GPUs he will be able to combine it with Nvidia's Grace central processing unit to create what he calls the GB200 superchip, allowing him to use large-scale language models. delivers up to 30x performance improvement on Nvidia H100 GPUs. The company says it will increase interfering workloads.
The system is named after David Harold Blackwell, a mathematician who specialized in game theory and statistics. Blackwell is the first black scholar to join the National Academy of Sciences, the company said.
Ian Buck, general manager and vice president of accelerated computing at Nvidia, explained in a briefing ahead of GTC: Rather than a single AI model, a collection of AI models called a mixture of expert models… These new models use multiple AI models and make them work together.
“Now, not only the queries and questions you ask the chatbot, but additional information fed from the database, from history and search engines, are fed into the model to create great answers,” Buck continued. “Rather than just taking a chip from Nvidia and putting it on a PC card, we're building a new kind of computer where multiple GPUs can work together to build bigger GPUs through a fabric called NVLink. We realized we needed a more capable system.”
Nvidia's DGX SuperPOD supercomputer system, also released at GTC, consists of eight or more DGX GB200 systems. These systems feature 36 GB200 superchips that work together as an integrated computer. The company said customers can expand the SuperPOD to accommodate tens of thousands of GB200 superchips depending on their requirements.
“The NVIDIA DGX AI supercomputer is the factory of the AI industrial revolution,” Huang said in a statement. “His new DGX SuperPOD combines the latest advances in Nvidia-accelerated computing, networking, and software to enable any company, industry, or country to improve and generate their own AI.”
At GTC, Nvidia also debuted Nvidia NIM-generated AI microservices, allowing developers to create and integrate customized AI applications while retaining intellectual property rights. We also announced healthcare-focused models and microservices to support drug discovery and drive digital transformation in healthcare. Additionally, a partnership between NVIDIA and Johnson & Johnson MedTech was announced focused on integrating generative AI technology into the operating room.
The company announced Project GRooT, a multimodal AI that will power humanoids in the future. This includes Jetson Thor robot computers with GPUs based on the Blackwell architecture and the DRIVE platform, which successfully connects the automotive industry to the creative AI era.
“The world is made for humans,” Rev Lebaredian, Nvidia's vice president of Omniverse and Simulation Technologies, said in a pre-GTC briefing. “General-purpose AI in the physical world will be humanoid.” Project GRooT includes the world's first robot that takes language, video, and human demonstrations as input and combines them with past experience to generate the next action. Contains basic models. ”
Nvidia also announced Isaac Manipulator, a platform with foundational models and acceleration libraries that will help make the world's robotic arms faster and more accurate, and an acceleration library and framework that will help make autonomous mobile robots smarter through 3D surround. We also announced our platform, Isaac Perceptor. , vision-based recognition.
“These smarter, faster, better robots will be introduced into the world's heavy industry,” Lebaredian said.
Finally, Nvidia announced the expansion of its Omniverse platform for developing and deploying physically-based industrial digitization applications (“a feedback loop for AI to enter the physical world”) across the global industrial ecosystem. . The company announced the Omniverse Cloud API, which allows developers to integrate core Omniverse technology into existing apps and workflows.
Omniverse allows industrial-focused organizations to use AI and physics-based 3D simulation to plan ahead before integrating robots into their facilities. This ensures that all possible scenarios are considered before actual construction begins.
Nvidia, a full-stack computing infrastructure company, was founded in 1993 and opened its Israel office in 2016.
In December, NVIDIA and its employees donated $15 million to Israeli and foreign NGOs supporting civilians affected by the October 7 Hamas massacre and the ongoing war in Gaza.
