Transcript

We use GPU's for a wide array of use cases, some of which, for example, simulations, GPU accelerated simulations, rendering use cases and of course model training and inference. And the thing I want to highlight and point out over here is that the use cases build off on each other. So we don't think of these use cases in isolation. An example here and we'll kind of hit on later in the talk is we rely on our bulk inference pipelines to be efficient because we use that data later on to train end to end models, right? And so a lot of the things that we're building. For front things on GPU build off of each other.