In an event that underscores a growing tradition between Nvidia and OpenAI, Nvidia CEO Jensen Huang personally delivered the first DGX H200 Artificial Intelligence supercomputer to OpenAI’s headquarters.
This gesture of partnership was captured in a photograph shared by Greg Brockman, president of OpenAI, featuring him alongside OpenAI’s chief Sam Altman and Huang, with the new DGX H200 prominently displayed.
Greg Brockman’s post on X highlighted the significance of this event,
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
This isn’t the first time Huang has played a direct role in providing cutting-edge technology to OpenAI. Back in 2016, during the early days of OpenAI, Huang delivered the first DGX-1 AI supercomputer, received by then active member Elon Musk, aiming to democratize AI technology.
Would like to thank @nvidia and Jensen for donating the first DGX-1 AI supercomputer to @OpenAI in support of democratizing AI technology
— Elon Musk (@elonmusk) August 9, 2016
The latest delivery comes at a critical time as OpenAI continues to push forward with its ambitious projects, including the development of GPT-5 and the upcoming public release of Sora.
The DGX H200, introduced by Nvidia last year as an upgrade over the highly successful H100, features significant enhancements that are set to boost OpenAI’s research capabilities.
With a 1.4 times increase in memory bandwidth and an 80% increase in memory capacity over its predecessor, the H200 is equipped to handle more demanding AI tasks effectively. Ian Buck, NVIDIA’s vice president of hyperscale and HPC, emphasized the H200’s role in advancing AI research,
He further explained.
Furthermore, the NVIDIA DGX H200 incorporates NVLink interconnect technology along with the NVLink Switch System, which allows 256 H200 superchips to be combined into a single GPU unit.
The tech enthusiasts even think that this might be used for some future GPT models,
Comment
byu/superman0123 from discussion
insingularity
This setup not only achieves an extraordinary 1 exaflop of performance but also offers 144 terabytes of shared memory, significantly enhancing the computing power available to OpenAI.
This collaboration between Nvidia and OpenAI not only demonstrates their mutual commitment to advancing AI technology but also ensures that OpenAI remains at the forefront of AI research with the most advanced tools at their disposal.
Tech enthusiasts are also appreciating this move,
They are doing meaningful things
— j (@jtsla4) April 24, 2024
As these technologies evolve, the partnership is poised to make significant contributions to the field of AI, potentially altering how we understand and interact with these emerging systems.
Found this interesting? Visit allaboutai.com for more daily exciting AI News like this.