Impossible to say based off of that.
The application might be single threaded, where A is a multiprocessor and one core is utilized heavily, and the rest of the cores sit idle and are less efficient. CPU B on the other hand could have less cores (Or even one theoretically) and be very efficient specifically for the type of process being done and be faster than the "faster" processor.
If CPU A did not support certain instruction sets, it might only use 30% of its overall power, but again a more efficient processor that handles the instruction sets might be faster at this task despite being a slower processor.
CPU A might have a higher clock rate, and more cores, but have a poor architecture. One possibility here would be if it was constantly making false predictions and having to clear / reload the pipeline. It might be a faster processor overall, but there always exists the fact that it will handle some tasks worse than a slower processor (Look back at AMD Barton cores vs the Netburst Pentium 4's, the AMDs were lower clocked significantly but on many tasks were faster).
CPU B could be better at shuttling data around from disk (SSD perhaps) and RAM, thus it spiked to 80% partially because it could load enough data to cache fast enough to fill up, while CPU A was on a 4,200 RPM laptop drive attached and burned each segment of data faster, but could not keep enough data in cache to keep up.
Now, none of this is definitive, but for the sake of the argument it just goes to show you there is WAY more to it then usage and raw measurements of speed.