should we expect 64 bit WIndows Phones very soon??? from the NT kernel??

[a5cent]

Contrary to popular belief:

1) The primary benefit of 64 bit processors isn't the ability to address more than 4GB of RAM, it's the additional registers(2x more and 2x wider than the 32-bit Cortex implementations currently in market) that the 64-bit ARM specifications require

Stop spreading disinfo.

It's a complicated topic.
There is nothing about a 64bit CPU that mandates it must have more registers than their 32bit brethren. It just so happens, that the ARMv8/A64 design has more registers. The way this is explained, at least 99% of the time, makes it sound like having more registers is a direct result of moving to 64bits, when that is not at all true. It's just that ARM decided to include both changes in this iteration of the CPU architecture spec.
Also note that we lack a clear definition of when a CPU is 32bit and when it is 64bit. Some processors claim 64bit after widening their address bus from 32bit to 40bits. On the other hand, the ARMv8/A64 design also includes FP registers that are 128bits wide, yet it sticks with the 64bit nomenclature.
From what I understand, it's precisely those 128bit registers, coupled with ISA extensions, which are responsible for the biggest performance gains, but only in very specific and limited circumstances. Graphics operations that run on the CPU would be one example, but this doesn't happen very often on a smartphone, as encoding/decoding is usually handled by specialized media processors and almost everything else should be done on the GPU. Almost anything related to encryption/decryption would be another example where wider registers are desirable, but ARMv8/A64 also includes new instructions specifically for that purpose. Those new instructions have a far greater impact on cryptographic performance than the 64bit'edness of those registers could ever have on their own. On the other hand, it's also true that pointer heavy operations are noticeably slower on ARMv8, by over 20% in extreme cases, as the CPU spends far more time moving pointers that "weigh" twice as much back and forth between CPU and memory.
My main points are these:
(a)
just widening registers doesn't always/automatically lead to improved performance. In general, the areas where improvements can be expected are few, at least on smartphones, while in other areas (even fewer) the move to 64bits can even be detrimental, and

(b)
while it's correct to say that 64bit ARM CPUs perform better than their 32bit brethren, far too many people attribute far to much of those improvements to the 64bit'edness alone. The direct consequences of that specific change are negligible, at least when compared to the other improvements made in ARMv8 CPU design. At least in the smartphone space, 64bit'edness, by itself, is still more hype than anything else. The ARMv8 spec would be more worthy of that hype, but you can't sell that to consumers.

 

[a5cent]

The answer to the OP's question is that we don't know, but I'd guess W10 at the earliest, so that means around mid 2015 or later.