On April 24, Nvidia's first superai chip, the DGX H200 system, was delivered by Nvidia founder and CEO Jen-hsun Huang, and OpenAI became the first user. OpenAI president and co-founder Gregory Brockmann posted the news on Platform X and posted a photo with Jen-Hsun Huang and OpenAI CEO Altmann.

Jen-hsun Huang's personal delivery of the DGX H200 to OpenAI is not only a symbolic gesture to strengthen the partnership, but also a testament to the critical role cutting-edge hardware plays in advancing AI technology. Jen-hsun Huang says it “Will advance AI, computing, and civilization”. As the two industry leaders continue to work together, the potential for artificial intelligence innovation is shown to be limitless and potentially transformative across industries. DGX H200 is Nvidia's latest and most advanced mass-produced AI chip, representing a major leap forward in artificial intelligence technology. It is also the most powerful AI chip hardware in the world. This delivery is a critical moment for OpenAI, a leader in artificial intelligence research. Compared with the previous generation H100, H200 performance improved significantly. Major upgrades include a 1.4-fold increase in memory bandwidth, a 1.8-fold increase in memory capacity, a total memory bandwidth of 4.8 TB/s, and a memory capacity of 141 GB.

These performance enhancements are driven primarily by the integration of HBM3E memory technology for faster processing and more efficient data processing. This is critical for training larger, more complex AI models, especially those of generated AI applications that generate new content such as text, images, and predictive analytics. In a recent presentation, Ian Buck, VP of high performance computing products at Nvidia, highlighted the performance of the DGX H200, “DGX H200's expanded and faster memory is designed to significantly improve performance for computationally intensive tasks, including training complex generative AI models and other high-performance computing applications, while optimizing GPU efficiency,” he said. For OpenAI, the acquisition of DGX H200 is an important strategic initiative that will enhance their AI research capabilities, in particular to drive the development of the highly anticipated GPT-5 large model. The increased capabilities of H200 will enable OpenAI to push the limits of what its AI models can currently achieve, particularly in terms of speed and complexity of data processing.

The impact of DGX H200 does not stop there. Its introduction into the market will advance the entire artificial intelligence industry, enabling researchers and developers to solve more problems, this will lead to major breakthroughs in areas such as drug discovery, climate modeling and self-driving car technology. High-tech industry has an unprecedented demand for high-performance AI processors, in this context, the supply and demand of H200 is also worthy of attention. H100 is in short supply because of its high demand. Nvidia is keen to avoid this when the H200 is delivered, by working with global systems manufacturers and cloud service providers. In response to a question about the high demand for and availability of gpus on a recent earnings call, Jen-hsun Huang said he would do his best to distribute them fairly and avoid wasting resources. But the actual delivery of the H200 remains to be seen.