AWS is continuing to offer services utilizing Nvidia’s Hopper chips, which are essential for training artificial intelligence systems. However, there is a shift towards using Nvidia’s new Blackwell chips, which promise increased performance. These Blackwell chips were introduced by Nvidia CEO Jensen Huang in March and are expected to be twice as powerful for training large language models like OpenAI’s ChatGPT compared to their predecessors.
The Nvidia GB200 Grace Blackwell Superchip integrates two NVIDIA B200 Tensor Core GPUs with the Nvidia Grace CPU via a 900 GBps ultra-low-power NVLink chip-to-chip interconnect. For optimal AI performance, systems powered by the GB200 chips can also be combined with the Nvidia Quantum-X800 InfiniBand and Spectrum-X800 Ethernet platforms for advanced networking capabilities at speeds up to 800 Gbps.
According to Neil Shah, vice president of research and Counterpoint Research partner, choosing Blackwell chips over Grace Hoppers makes more sense for advancing language models and simulation AI training across various industries. He emphasizes the importance of AWS staying at the forefront in the AI race, especially with the rapid evolution in the size of training models and the need for efficient computing.
Major cloud service providers like AWS, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure will be among the first to offer instances powered by Blackwell chips, as announced by Nvidia in March. Additionally, companies in the Nvidia Cloud Partner program, including Applied Digital, CoreWeave, Crusoe, IBM Cloud, and Lambda, will also provide these advanced instances.
In conclusion, AWS and other cloud service providers are transitioning towards using Nvidia’s new Blackwell chips for improved performance in training artificial intelligence systems. The advancement in chip technology is crucial for meeting the evolving demands of AI applications across various industries. It is important for AWS to stay at the forefront of this technology race to remain competitive with other major players like Google and Microsoft. By investing in advanced computing infrastructure and optimizing energy consumption, AWS can achieve the best return on investment and efficiently keep up with the pace of innovation in AI technology.
Article Source
https://www.networkworld.com/article/2118947/aws-clarifies-nvidia-chip-order-upgrade-not-halt.html