I mean, it might have prevented learning about and using new undocumented APIs going forward, but it would be in none of the newly-broken-up companies' interests to break the existing usage.
> splitting some of those divisions further into competitors
Even that I'm skeptical of - first, because you'd need enforcement to prevent them from simply merging again after a while (glares at Ma Bell breakup companies reforming into a few huge companies which usually deliberately avoid competing in markets), and second, I'm not convinced that deliberately fragmenting Windows into 2 distinct codebases is beneficial for consumers?
Even assuming you could avoid one of the two competitors simply winning the vast majority of the marketshare after one round of OS upgrades, you'd likely end up with mutually incompatible API surfaces, thus breaking one of the main reasons people like Windows - compatibility.
(I still claim, though, that given two initially equal products and equal resources, that one will probably relatively quickly eat the other's lunch. Look at what's happened almost every time a large open source project has had a high-profile fork - look at egcs and gcc, or OpenWRT and LEDE, or libav and ffmpeg.)
That kind of breakup wouldn't have solved that anti-competitive problem, but it would have solved others (e.g. the Office Team being able to use undocumented OS APIs to get an edge on competitors).
Though, personally I kinda think a breakup of a tech monopoly should be more extreme, for instance by breaking it up by division and splitting some of those divisions further into competitors (e.g. split the OS division in two, each with their own Windows fork to sell, and all the Windows trademarks go to some independent interoperability consortium).