1. Geekbench doesn't mean a damn, and unfairly advantaged for ARM architecture based on AES/SHA acceleration 2. Apple doesn't run Windows, so A8X performance doesn't mean a damn, unless you are comparing an iPad Air 2 against an equivalent tablet 3. When Intel needs to add another core or two to Core M to win in performance, they will, and wipe the floor in MT. In the meantime, Core M is a smaller die and even less expensive to make than A8X. 4. "More fairly priced" is your strawman. They are already fairly priced, because they sell at those prices. If there were already an alternative threatening to be a better Windows platform, then the pricing and sku definition would already be different.
1. Skylake is less than 6 months away from volume availability. Volume production starting very soon.
2. 30% performance improvement and Intel official claim of 60% less power, exceed the improvement from Samsung process improvement going from 20nm to 16nm Finfet.
3. Intel's costs for 16nm are declining pretty rapidly. By all accounts Samsung 16nm will be higher cost per transistor than Samsung 20nm even at mature production which is not the case for either at this point.
In a nut shell, the Sky Lake design, which is likely before any Apple A9, provides more than the benefit of Apple moving to Samsung 16nm and at declining rather than rising costs. Of course Apple may provide improvements over the A8 but at the cost of even higher cost transistors.