“We selected Oracle because of the affordability and performance of the GPUs combined with Oracle’s extensive cloud footprint. GPUs are very important for training deep neural network models. The higher the GPU performance, the better our models. And because we work in several different countries and regions, we needed the …

“Our training and inference models are resource-intensive and need powerful GPU and CPU instances. A typical virtual machine requires software that includes a 10–15% performance tax. GPUs don’t require that underlying software. Our processes run anywhere from 24–30 hours at each stage, so even a 10% improvement in performance means …