In a significant milestone for the advancement of artificial intelligence, NVIDIA CEO Jensen Huang personally delivered the first NVIDIA DGX H200 to OpenAI. The momentous occasion was shared by Greg Brockman, president of OpenAI, in a post on X (formerly Twitter). The image shows Huang posing with Brockman and OpenAI chief Sam Altman, with the DGX H200 taking center stage.
First @NVIDIA DGX H200 in the world, hand-delivered to OpenAI and dedicated by Jensen “to advance AI, computing, and humanity”: pic.twitter.com/rEJu7OTNGT
— Greg Brockman (@gdb) April 24, 2024
Brockman captioned the post, “First @NVIDIA DGX H200 in the world, hand-delivered to OpenAI and dedicated by Jensen ‘to advance AI, computing, and humanity’.” This gesture seems to have become a tradition, as Huang had previously donated the first DGX-1 AI supercomputer to OpenAI in 2016, which was received by Elon Musk, in support of democratizing AI technology.
The Upgraded DGX H200: Enhancing Performance for Generative AI
The DGX H200, which succeeds the highly sought-after H100, boasts impressive upgrades that could prove invaluable to OpenAI’s current projects, including the development of GPT-5 and the planned public release of Sora this year. The H200 features 1.4 times more memory bandwidth and 1.8 times more memory capacity compared to its predecessor, significantly enhancing its ability to handle demanding generative AI tasks.
The H200 also introduces a faster memory specification called HBM3e, which increases its memory bandwidth to 4.8 terabytes per second from the H100’s 3.35 terabytes per second. Additionally, the total memory capacity has been increased to 141GB, a substantial increase from the 80GB offered by the previous generation.
Ian Buck, vice president of hyperscale and HPC at NVIDIA, emphasized the importance of these upgrades, stating, “To create intelligence with generative AI and HPC applications, vast amounts of data must be efficiently processed at high speed using large, fast GPU memory. With NVIDIA H200, the industry’s leading end-to-end AI supercomputing platform just got faster to solve some of the world’s most important challenges.”
NVIDIA Launches New AI Supercomputer with H200
Alongside the delivery of the DGX H200 to OpenAI, NVIDIA also launched a new AI supercomputer utilizing the H200 GPU. The NVIDIA DGX H200 harnesses NVLink interconnect technology and the NVLink Switch System to combine 256 H200 superchips into a single GPU unit. This configuration achieves an impressive 1 exaflop of performance and offers 144 terabytes of shared memory, representing a significant leap from the previous generation NVIDIA DGX A100 introduced in 2020, which had considerably less memory.
The Future of AI and Computing
The delivery of the first DGX H200 to OpenAI and the launch of the new AI supercomputer demonstrate NVIDIA’s commitment to advancing AI technology and supporting organizations at the forefront of AI research and development. As OpenAI continues to push the boundaries of what is possible with AI, the enhanced capabilities of the DGX H200 could prove instrumental in achieving new breakthroughs and unlocking the full potential of generative AI.
With the dedication of industry leaders like Jensen Huang and the ongoing collaboration between NVIDIA and OpenAI, the future of AI and computing looks brighter than ever. As these powerful tools continue to evolve and become more accessible, the potential for AI to transform various aspects of our lives and address some of the world’s most pressing challenges grows exponentially.
