It seems that somehow, the Sailrocket team has found a way to delay cavitation to a much higher speed than what some of the theories have predicted in the past.
Please allow me to ramble a bit. Just thinking out loud really. I could be way off.
I can speculate as to what they might have done. In theory, to delay cavitation you have to reduce the peak load on the foil so that the low pressure side does not get vaporization (boiling). I can also speculate about some ways on how to do this.
-Use a larger foil to reduce the lift per unit area. (but increases drag)
-Use a longer chord, lower aspect. (but reduces lift to drag ratio)
-Design the foil section to have a more even lift spread over the whole section, rather than concentrated on one part. (I am sure there could be any number of negative effects of this but I don't know enough to comment further)
-reduce the angle of attack and total load on the fin (Windsurfers may do this to some extent by sailing broader, but SR2 is sailing quite a tight angle in comparison)
I am sure there are others and someone will add to this list.
Note that all these directions have been tried in windsurfer speed fins and are currently evident to some degree on the fins used by the fastest windsurfers.
The problem as I see it, is that much of this is theory. The computer programs that designers use to predict cavitation of different foils are not absolute. They are based on best knowledge, but that is not absolute. There seems to be a large number of variables that the programs can not always accurately predict. Its the old computer truism again: The data out is only as good as the data in.
The fascinating thing about Sailrocket is that it seems to have added a significant new chapter to the book. I can't wait to read that book.

On the other hand, when Hydropter did it's thing and amazed us all a couple of years ago, we had a similar conversation. Mal Wrights take on it was that, perhaps there was significant cavitation happening, but that Hydropter could develop so much power it could overcome the extra drag and still go faster. Perhaps this is part of the equation with SR2 as well?