NVIDIA To Develop 400 Petaflop DGX A100 Supercomputer For UK AI Research
The Cambridge-1 supercomputer will be an NVIDIA DGX SuperPOD supercomputer with over 400 petaflops of AI performance and eight petaflops of Linpack performance. This sort of horsepower would allow it to rank 29th on the TOP500 list of most powerful supercomputers in the world. Inside, it will feature 80 modular NVIDIA DGX A100 systems connected by NVIDIA Mellanox InfiniBand networking. Also, the modularity “enables the system to be installed and operational in as little as a few weeks.” When it is completed, NVIDIA will have invested approximately $51.7 million into the system.

The research NVIDIA is driving with the Cambridge-1 supercomputer is set to be the first in a long line of future projects. After announcing its plans to acquire Arm, NVIDIA intends to develop an “AI Center of Excellence” at the ARM headquarters. Among other supercomputers, this new center will feature an Arm-based supercomputer. Eventually, Cambridge-1 will also be brought into the center, which will expand in the future.
With these significant investments in the field, NVIDIA is driving the world of AI at present. As it stands, healthcare is the current focus, but that will expand as the AI Center of Excellence grows. Once these developments are completed, NVIDIA will be a company to watch as they help to improve the world around us by harnessing and democratizing AI.