1) 2048 qbits doesn't help you crack 4096 bit keys. Actually, it's dubious it will help you crack 2048-bit keys. I remain confident the NSA does not have a quantum computer with that capability because they just aren't that well funded.
You see, in order to get a quantum computer like that, they'd have had to have developed it outside the mainstream of academia and industry. We're missing all sorts of precursor technologies to get us there, which means that to assume the NSA have a quantum computer that can crack hard crypto you have to assume they managed to get enough of the right people to develop it in total secrecy without ever publishing a paper on it. Have you met the kinds of people who have the ability to do that kind of research? Convincing them not to publish a paper on it is damned near impossible.
2) As I said in my previous post, I am pretty sure the proper hard crypto algorthims remain uncompromising simply because so many people have attacked them for so long. These are not algorithms that were developed in secret and that remain secret. (Bitlocker, as one example.) These are public knowledge and the best minds in the world are constantly trying to break them. So far, with little success.
3) Explain to me how you feel you can "compromise" hardware processors in such a way that they specifically create a back door in any cryptographic algorithm they generate? This might be possible with specialist chips like TPM, but a general purpose CPU or GPU? Do you honestly believe the CHinese wouldn't have found that by now and exploited the living piss out of it?
4) Windows has all sorts of backdoors. Bitlocker is a great example. If you don't use the operating system's libraries to generate your crypto you're fine...or are you going to tell me that suddenly there's magic voodoo within Windows that has heretofore gone unnoticed that simply "knows" (how?) when a library or thread is running "some form of cryptography" and magically backdoors it?
A crypto library that Microsoft ships as part of their OS certainly can be compromised. They probably all are. TPM is probably completely untrustworthy as well. ("Trusted platform module" my fat, jiggly ASCII.)
But the generic computing stuff? CPUs, GPUs, basic execution of libraries written by third parties? To compromise that? We're not talking about your run of the mill engineer here. We're talking about potentially requiring the single smartest individual the human race has ever produced. Someone who would be able to learn so much - to know and retain so much - about how so many different things worked that he would make Leonardo Da Vinci look like Honey Boo Boo.
If such a person existed and were identified by the United States Government before anyone else...do you honestly believe - really and truly, deep down in your heart of hearts - that they would waste that person's talent so utterly by having that individual come up with new and interesting ways to compromise cryptography in generic computing systems?
Why? Where's the logic in that? What possible reason could they have for that when there are way easier methods available? Man in the middle fibre taps. $5 wrench. Secret letters demanding keys from providers and crypto implementers.
I think your tinfoil hat is on too tight, buddy. You should visit my guy. He custom-manufactures mine and it's quite comfortable to wear.