Heating tin in a regular laser until it forms plasma and starts emitting 13.5 nm light is much simpler to scale down (since it doesn't involve an accelerated electron beam) but you lose large amounts of power while trying to form a partially coherent beam out of the result, turning hundreds of kilowatts of infra-red laser power into mere 10's of watts of EUV power.
The high input to output ratio means that you need to be really careful that the large amount of dissipated power doesn't cause problems. Tiny misalignment problems often result in the 200kW infra-red input laser melting the entire device into a puddle of former components.
http://www.eetimes.com/document.asp?doc_id=1321162&page_numb...
There are interesting ideas to downsize synchrotrons though, and if we can both make synchrotrons smaller and semiconductor factories bigger, then perhaps synchrotron sources could make sense. But even then, I worry that if a $20 billion semiconductor plant has only one $1 billion synchrotron source, there is a tremendous amount of risk concentrated in that single source. If the source goes down, the rest of the $20 billion plant goes idle.