Great, tell me those tons of reasons. I can imagine the surveillance tech is somewhat better developed, but I can't think of any feature that one wouldn't be able to implement on Linux and even then end point security is not really the way security is going anyway.

If those hardware backdoors exist, then why does the NSA need to intercept routers in the mail to modify them? If you say it's just in the CPU, then why not just run an FPGA with a softcore? If you are saying the government has super technology capable of recognizing a custom instruction set on such an FPGA, then you don't understand the fundamental limitations of technology.

Even ignoring all that, one could certainly build a router that terminates when a single packet is received that is unexpected.

Now, I do think that access to some random Windows system is probably quite easy, but getting access remotely just doesn't seem to be possible (hardware vendors such as Intel also claim there are no designed backdoors, btw, which means that they could be sued if it turned out to be otherwise). (Having said that, the design of any high performance chip is insecure since SPECTRE). Take the attack on Iran, which required a USB-key to get access to some maintenance laptop, not Internet.

All the old chips that have been reverse-engineered do not contain a government backdoor.

Imaging technologies are so good these days that it should be possible to X-ray every transistor, especially on larger nodes and produce a secure device. I'd expect that Intel, AMD, Apple devices are all compromised in some way, but I just don't see some mass produced microcontroller to have such backdoors, similarly for risc-v devices produced in e.g. China.