Even so, any wafer those "D" came from will not have grade "A" if A means going that much faster than what these are doing.

Besides, the logic of this exercise this escapes me somewhat. BFL must have thousands of these chips, surely using TWO grade A chips isn't going to make a lick of difference in terms of supply?
Which school of business did the decision makers study to use the worst hardware they have on the shelf and then use that to demonstrate the functionality of an upcoming product? If they have A grade chips they would be able to clock them way higher without actually drawing more power through the PCB. And without a hilarious error rate. Having a board hashing at slightly more proper speed and with a low error rate might stem the flow of refund requests a little, at least. Seems worth using a tiny fraction of a percentage of A chips.

But, according to you, they don't want to waste any of the "good" chips on a demo. They must really be short on A grade chips. Do they even have two?
Any info from BFL is difficult to take seriously these days, but if it makes as little sense as this I'm confident to call it nothing but pure undiluted BS.