Jay, you opened a very big box with silicon lifecycle management because there are multiple entry points to that and one of the entry points that sort of obvious and actually have been asked for many years, is for chips that are in the midst of a situation that could be life threatening i.e., in charge of a car or the brake system of a car, so you really would like to know, is the chip still working. And if the answer is no, you really want to know what to do with that answer or how do you stop the car or whatever you have to do. And so that is already the first example where for a number of years, people have started to put different mechanism inside of a chip via test connections to know is something is broken, is the software still running correctly or are there heat situations and so on. And so from there to go to the notion of, well, why don't we start to add some additional sensors and by the way, sensors are very little things. So you can easily find a corner somewhere to put them in, that brings very quickly the next question, well, why don't we then take some data inside of the chip and can we make the decision of the chips that works inside of it, with other words can you have some machine learning that gets interpreted inside of the chip. In contrast to say, all we can now read out a lot of this data from the chip that comes from these sensors and use external learning or external interpretation of the data, so that we can make better chips going forward. And so I'm trying to just give you a little bit of a visceral sense that the minute you put observation points in anything but certainly on the inside of a chip and you have a way to either bring the data out or use it on site, so to speak, all kinds of new doors open up and that is why this is exciting, because on one hand, you could say, well, nothing new here on the other hand, you'd say there is a lot of new here and a lot new opportunity and specifically the AI that can be applied with it. So hopefully that gives you little bit of a sense that the opportunity space is very big. The good news is we are in a leading position with our test capability, so we have good connectors there, we have a very strong machine learning capabilities, we can use that and by the way our own IP has already use a number of these type of mechanisms for a number of years. So technically we know what to do.