May never ROI. Since difficulty can fall (or stop rising), there are infinite scenarios under which a BFL Single could generate ROI.
However, the more BTC rises in USD, the better off the early pre-order people would have been just buying BTC from Gox or BTC-E instead of BFL.
In almost every case, buying bitcoins would have been smarter, easier and more lucrative than buying ASIC mining hardware.
That is a separate subject from whether or not my BFL hardware returns the initial investment (it will).
There are only 3 "cases" of ASIC hardware so far. Avalon batch #1 was better than just buying BTC, so were early purchases of ASICMiner shares.
When you bought a BFL unit, you bought a variable return of Bitcoin that arrives in the form of hardware. The comparison you should be making is "is this variable return greater than the fixed return of a straight BTC purchase." You should be using that as the yardstick because you do not control the exchange rate of Bitcoin. It could drop to $1/BTC tomorrow and you would not earn a USD return on your initial investment. Obviously, every purchase of either mining equipment or BTC is an investment in bitcoin itself and runs the same risk.
Example: If one buys gold, and gold rises, then accolades are in order. Unless one bought the gold in the form of numismatic collector coins, then accolades are not in order. Even if gold doubles in price, it is unlikely for that collector to recoup his initial investment. Numismatic coins are a bad way to invest in gold. Eventually, gold may triple in price and the investor may recoup his initial investment, but it is not a happy ending.