Yeah this looks like an SOA violation to me based on the hole. Usually you would turn a FET like that on slowly to avoid high inrush current into the downstream

Yeah this looks like an SOA violation to me based on the hole. Usually you would turn a FET like that on slowly to avoid high inrush current into the downstream circuit, but ~10ms seems a bit silly. That would also explain why this escaped their initial testing because the FET wouldn't get hot in normal operation. Although, if it's a 5V0 or 3V3 rail they may just not be turning on the FET enough. It's Rds(on) is high at ~3 V, and they divide down the gate drive to get reasonably symmetric on/off times and slow speed.