With Apple hardware generally being slower and more expensive, sometimes the superiority of the SDK isn't the most critical element.

For some tasks, the Apple code is orders of magnitude faster so it's not a problem. The ability of the Mac Pro to smash through H.264 video with GPU acceleration is pretty much unparalleled. Compared to FFMPEG it's not even close. When it comes to general compute, though, it's easily outclassed.

It's all about benchmarks and cost/benefit analysis.

I've seen code that works on the iOS 20-100x faster than a desktop equivalent because the iOS version is done using Metal and the desktop one is barely vectorized CPU code. In some edge cases a "server farm" of iPads might outperform an equivalent spend on Xeon-based servers.