OS/2 3.0 "Warp" was a little too much ahead of its time and had the exact same problem that Windows Mobile had: no applications.
IBM tried to solve that with Windows emulation but it was a headache from the start and often have a buggy experience.
It didn't help that the real world hardest requirements were off the charts as compared to Windows 95 (still 16-bit MS-Dos based and not even close to what OS/2 was).
IBM did everything right from an engineering perspective but failed miserably on what the market wanted.
It never stood a chance. IBM had always been great at delivering solutions that was well engineered. What IBM has n-e-v-e-r been good at is marketing and understanding the volume market.