These asics have own risks - if difficulty starts to grow exponentially, as with bitcoins, they'll never going to get to break-even.

Even with current difficulty it will take almost 7 months to get to break-even (http://bitcoinwisdom.com/litecoin/calculator), but guess what happens when they start delivery en mass and prices drops too? And right now it looks like some sort of school project - no documentation for device that potentially overtakes a hundred mil GPUs market for scrypt? Either somebody doesn't want to give away information, or knows something that better not to document.

So difficulty should not exceed 7000 by year end to get devices to break-even (if exchange rate stays the same of course).
Or exchange rate should continue to grow along with difficulty (best scenario), but unfortunately there is no way to hedge against changes in exchange rate.
It would be nice to be able to sell bunch of futures contracts on LTC, but none available afaik

GPUs, on other hand, won't be worthless even if whole things collapses tomorrow.

There are benefits, of course - electricity bill, maintenance supposedly much better. But I think changes are close to zero there will be second-hand market for these devices