Prediction: over time, there will remain no commercially significant high-end chips without this sort of a "security subsystem." Recommending ARM, PowerPC or other architectures over x86 as the FSF suggests will not get you very far because the problem (or as chip vendors call it, the solution) is not in the ISA, it's in the chip, and every chip vendor making the sort of chip that can power a general-purpose computer will end up being this way.

Of course in practice, as long as say Linux runs on the machine, the existence of the ME or the like is almost inconsequential for the user, because you run the same software, and there's never a shortage of security vulnerabilities right there in the OS and the userspace software. In terms of the impact on the number of vulnerabilities, eliminating C and C++ would go way further than eliminating black box "security" hardware, just because the huge amount of C and C++ code, much of it written hastily and committed the moment it "runs on my machine", presents a much larger attack surface than the black box hardware + software system. But of course the FSF will never recommend ditching C and C++.