I've been puzzled by this as well. My suspicion is that chip ampacity is more a figure of tolerance to heat & thermal conductive latency than raw power wielding.
I'm thinking a larger ampacity die will handle more switching current heat and allow the package to "catch up" before internally burning up. But not as a substitute for spreading the average load across enough heat sunk area.
I've only played in power stuff, and willingly yield to my betters... <<<)))