NVIDIA claim that advancements in AI and accelerated computing are leading to substantial energy efficiency improvements in various sectors, from finance to manufacturing, challenging misconceptions about AI's energy consumption.
Nvidia state that Accelerated computing using NVIDIA GPUs consumes less energy than CPU-only systems for complex tasks, especially in AI workloads. Users across various sectors are reporting substantial energy savings. For instance, Murex, a financial services company, saw a 4x reduction in energy consumption using NVIDIA Grace Hopper Superchip. Wistron, a manufacturer, increased overall energy efficiency by up to 10% using NVIDIA Omniverse.
In scientific research, the National Energy Research Scientific Computing Center, found that applications accelerated with NVIDIA A100 GPUs, saw energy efficiency rise 5x on average. NVIDIA-powered systems also dominated the recent Green500 ranking of the world's most energy-efficient supercomputers.
The company is also working on innovative cooling technologies for data centers, aiming to improve efficiency by 20% over current air-cooled approaches.