How to crawl billions of images from all around the web

While the product this is content marketing for (Mixnode) looks interesting I was hoping for more technical details given the title.