The reason is that they want to see evidence that you can take an unstructured problem where you don't have the faintest clue where to start, and make enough forward progress on it, on your own, that some useful contribution to human knowledge will come out the other end. This is a very different skillset from most software engineering jobs - when you're developing new knowledge, Google is useless, StackOverflow is useless, your peers are mostly useless, and you need to know how to ask the right questions and find the answers to them. A Ph.D is an institutionally-sanctioned way of demonstrating this skillset. There are other ways to demonstrate it - work in an industry research project that's successful and perceived as cutting-edge, founding a technology-based startup, published independent research that's cited by others - but they all involve you taking on the risk of conducting your first major research project on your own, without institutional guidance.