In 2017, there was a paper that came out called attention is all you need about a new type of machine learning. It understands principle similar to the human brain because we tend to filter information and pay attention to what's important. So with stable diffusion, our image model that we released, we took 100,000 gigabytes of images and the labels of those images. How many images is that or do we know? Two billion. That is not actually compression because it's impossible to compress that amount unless we invented any type of compression in which case, hey, my company's worth a trillion dollars now. Just line up. Instead of what it did, it was it learnt the principle.