CARRIER x PN mod. BPSK = Spread Spectrum? How?

I am trying to understand the very basics of spread spectrum systems (I know pretty much nothing in RF..) and I need some help. I could understand the idea, I find it very exciting, but the Direct-Sequence Spread Spectrum is now confusing me. I found several doc's saying that commonly it is used with binary Phase-shift keying modulation (BPSK). So, to spread the spectrum, we must multiply the carier signal with a PN code, and then modulate this with BPSK. I really dont understand why the spectrum is narrow if we direct modulate the carrier, and it spreads if we have multitply the carier before with a PN code. I could understand if we are saying that the final signal frequency maybe goes higher (because of the higher carier frequency) , but the spectrum spreads? A carier of higher frequency (to my understanding CARRIER x PN) when modulated with BPSK, spreads the spectrum? Why? It makes no sense to me. My brain is stuck

*(FHSS is more simple, and in general i think I understand it)

The PN code is much faster than the data bits. The best way to implement this is to XOR the data with the PN code at the digital baseband level. Usually the PN code is an integer multiple of the data rate.

Ok, so a PN code XORed with the baseband signal results in a much higher frequency pseudorandom "whitenoise-like" (binary) signal witch then goes for modulation eg BPSK. I think that this means that much more frequent "phase shifts" in the transmited signal will take place. But continue to not understanding how this spreads the spectrum (ie extends the bandwidth) of the transmited signal.