Gridseeds will likely have a positive ROI for some time. With multipools along with multiple lucrative coins, it's going to take awhile to get enough hashrate online to saturate everything to the point where the power sipping gridseed is no longer profitable to run. At average US power rates, LiteCoin would need a difficulty greater than 50,000 before it became unprofitable to run a Gridseed.
The problem is that ROI is positive return after initial purchase price. So say I spend $4000, I need to make at minimum this amount back +1 for a positive roi. Litecoin for example is currently at ~5000 difficulty so we would need a 10x increase to reach a point where it is unprofitable to turn on your asic. We wont see that in the next few months but we will see a consistent increase in difficulty thus constantly postponing the date when we see positive roi.
The problem is if new asics are 100x more powerful than current gridseeds, could we reach 50,000 difficulty before a positive roi on gridseed miners? And his goes for more profitable altcoin mining as well.