NVIDIA announced its next-generation Blackwell GPU architecture, designed to catalyze a new era of accelerated computing and enable organizations to build and run real-time generative AI on large-scale language models with trillions of parameters.
The Blackwell platform reduces cost and energy consumption by up to 25x compared to its predecessor, the Hopper architecture. Named after pioneering mathematician and statistician David Harold Blackwell, this new GPU architecture introduces six groundbreaking technologies:
“Generative AI is the defining technology of our time, and Blackwell is a driving force behind this new industrial revolution,” said Jensen Huang, founder and CEO of NVIDIA. “By working with the world's most dynamic companies, we are realizing the potential of AI across every industry.”
Key innovations in Blackwell include the world's most powerful chip with 208 billion transistors, a second-generation Transformer Engine supporting twice the compute and model size, fifth-generation NVLink interconnect for high-speed multi-GPU communication, and advanced engines for reliability, security and data decompression.
At the heart of Blackwell is the NVIDIA GB200 Grace Blackwell superchip, which combines two B200 Tensor Core GPUs and Grace CPUs with an ultra-fast 900GB/sec NVLink interconnect. Multiple GB200 superchips can be combined into systems such as the liquid-cooled GB200 NVL72 platform with up to 72 Blackwell GPUs and 36 Grace CPUs, delivering 1.4 exaflops of AI performance.
NVIDIA has already secured support from major cloud providers, including Amazon Web Services, Google Cloud, Microsoft Azure and Oracle Cloud Infrastructure, which are offering Blackwell-powered instances. Additional partners planning Blackwell products include Dell Technologies, Meta, Microsoft, OpenAI, Oracle, Tesla and many others across hardware, software and sovereign cloud.
“We are fortunate to have a long-standing partnership with NVIDIA and look forward to bringing the groundbreaking capabilities of Blackwell GPUs to Google's cloud customers and teams to accelerate the discoveries of the future,” said Sundar Pichai, CEO of Alphabet and Google.
The Blackwell architecture and its supporting software stack will enable new breakthroughs across industries, from engineering and chip design to scientific computing and generative AI.
Mark Zuckerberg, founder and CEO of Meta, commented: “AI is already powering everything from large-scale language models to content recommendations, advertising and safety systems, and it will only become more important in the future.
“We look forward to using NVIDIA's Blackwell to train our open source Llama models and build the next generation of Meta AI and consumer products.”
With its massive performance gains and efficiency, Blackwell could be the engine that finally enables real-time trillion-parameter AI for enterprises.
reference: Elon Musk's xAI open-sources Grok
Want to learn more about AI and big data from industry leaders? Check out the AI & Big Data Expo in Amsterdam, California and London, a comprehensive event taking place alongside other major events like BlockX, Digital Transformation Week and Cyber Security & Cloud Expo.
Find out about upcoming enterprise technology events and webinars hosted by TechForge here.
