![[CROSS]](../cross.gif)
Windows NT has Microsoft behind it, and it will be pushed onto desktop computers regardless of whether it's wanted or needed. And it probably will be considered worth it. It already has subsystems to cope with binaries from OS/2; it has Posix which might be expanded to include Linux. And ARDI's Executor already exists to run Mac programs. NT is already available on a good variety of platforms.
IBM has put far too much effort into OS/2 to let it drop - 10 years and three billion dollars. And a lot of people consider it to be one of the best systems available. If one also includes ATM machines, and all the various "ordinary" devices which we interact with, most of which run OS/2, the 3% figure looks pretty easy to acheive. OS/2 is being microkerneled, and will probably be expanded to run Win-32 binaries. It already has a Posix subsystem, and IBM have licenced MacOS already. (Not specifically for OS/2, merely for distribution, but the next step isn't that far.)
MacOS will probably be around for ages to come - Apple have such fantastic brand loyalty that it's hard to imagine them going away in the near future. SoftWindows is already available, and can run i386 binaries for any operating system already. (But slowly).
Linux is growing staggeringly fast, and has shown itself to be a workable development platform, and a likeable system, even for non-hackers. Wine will be finished by then, and if there is enough interest, an OS/2 emulator will also exist - possibly also written by ARDI. The Executor is already available. Ports are already underway, (or completed) to all the above platforms.
Who knows what the other one will be? It might be Taos (possible), it might be the Hurd (maybe), it might be Tunes (hmmm). Almost certainly it will be freely available - only the very biggest of companies nowadays have the resources to build a complete operating system from scratch, and it is almost inconceivable for a new operating system to return a profit in any reasonable length of time.