Wavecom puts Lua on its embedded wireless devices (from 256KB to 1MB RAM, with a pretty constrained OS underneath), and uses it among others for all their web server stuff. http://www.wavecom.com/openatlua. In a non-realtime embedded context, coroutines are a real blessing, giving you most of multithreading's advantages (arguably more, since getting race conditions right is much easier) for a fraction of the resources' cost.