Regardless of bloatware, I don't see how more transistors can indefinitely simulate an increase in processor speed. It seems like it would violate P!=NP or the

Regardless of bloatware, I don't see how more transistors can indefinitely simulate an increase in processor speed. It seems like it would violate P!=NP or the halting problem or something - as well as being a massive security hole. This is why gpgpu processing seems like the wave of the future. Alternatively, human level AI won't come from OpenAI but the final desperate Intel team trying to create a system that anticipates all serial routines the processor can run in parallel.