Instead I think the best goal is the algorithm should be one that it is likely to be highly needed for general computing and thus Intel will continue to accelerate it and motherboard vendors might even build on an ASIC to each motherboard. Bitcoin USB ASICs with the similar power efficiency of the bigger models now cost only $15 at Amazon.com.
Isn't Bitcoins SHA256 used widely not only for mining already?
Intel is proposing to accelerate SHA2. But this is going to take years to be implemented and adopted by a majority of computers.
AES is already accelerated on Intel CPUs. And every one is going to need encryption in this brave new NSA world.
I'd bet on AES instead of SHA2 as reaching ubiquity sooner.