Gary E. Muenster
Management
That's a right way to look at it, Kevin. If you look at -- let me give you 2 data points. One is our content for plane. Obviously, A350 is the largest contribution per aircraft, and that's about $130,000 per unit on the OEM side. And what they're publishing in all the airline monitor and all the other things that come out of the TL report for directional guidance there, is they expect that to get up to about 110 planes a year by 2017. And so sitting here where we are today, obviously it doesn't go from 0 to 110, just step function. So I think if you model that out going up to 50 planes a year, next year, and 75 or 80 planes in 2016, and then put yourself up to the 117. So that's kind of the glide path of the development of the revenue profile there. And when you look at some of the other things, the next one I'll talk about is the -- we announced was the COMAC, which is the Chinese C919. Our content on that is about $10,000 a plane. And that really kicks in, in earnest, in 2017 as well. The Embraer, the E2, the ERJ upgrade we're talking about, we do about $40,000 per aircraft on the OEM side of that. And that kicks off in earnest in late '17, early '18. And the build rates they're looking at there are in the high double digits, $50,000 to $90,000 a plane. I think stepping into '15 and '16 you're going to see nice organic growth. And in 2017, when all those programs are running on what they think is close to annual stable run rates, I think that's where you'll see the step function change.
Kevin R. Maczka - BB&T Capital Markets, Research Division: Got it. That's great, Gary. And then in terms of the engineering and new product development costs associated with these programs and others that you're bidding on, and hope to win, I guess for the existing ones that you've already won, are there some engineering costs that roll off and go away next year? Or is this just kind of part of doing business from here on out, as you're trying to develop new products and win new programs?