Take a close look at the chart below (from this article), which shows the market share of operating systems as measured by the number of “computing platforms” (personal computers, laptops, tablets, smart phones, etc.) that use them. Note that this same graph using a metric of operating system revenue or profit would look very, very different (for instance, Android would be zero, since it's free). There's also something else odd here: Linux is missing except for Android.
One thing that jumped out at me is very personal: this chart almost exactly captures the range of time of my own interest and involvement with personal computing platforms: I first built a “personal computer” in 1974, just one year before the beginning of this chart. That year would have been 100% “other”, because the only commercial personal computers were offerings like the MITS Altair; brands that never took off. Most personal computers then were home-built, as my first few were.
The big story here is the rise and fall of “WinTel” (Intel chips running Windows). I'm not sure precisely how these WinTel numbers were generated, but they must include MS-DOS computers as well, because Windows certainly didn't have significant market share in 1985, let alone 50% (Windows really didn't take off until the release of Windows 3.0/3.1 in 1990). But this is a quibble - rename it “Microsoft and Intel” and it's clear. Two or three years ago, the WinTel share slipped below 50% again, for the first time in nearly 30 years. In the technology world, that's an extraordinary run. Furthermore, if this graph were in operating system revenue instead of device count, WinTel would still be over 50%. Amazing!
But the device count for Android and Apple matter. Android is entirely smartphone and tablet, and the Apple count is dominated by IOS (its operating system for portable devices). The Mac OS/X count is tiny by comparison. All these devices require applications, and the ecosystem of application developers is rapidly growing for portable devices. I'm not sure the traditional application development world is shrinking – I suspect, in fact, that it's still growing. It's just that the portable device application development world is growing much faster yet. Why does this matter? The success of those applications is what really drives the operating system penetration. Apple and Android are perceived as “sexy” platforms to write for right now, and Windows is perceived by most developers as “legacy” and, well, stodgy.