You are correct, at least as far as current theory and technology is concerned. There must be a conventional carrier signal and the actual latency is the latenc

You are correct, at least as far as current theory and technology is concerned. There must be a conventional carrier signal and the actual latency is the latency of that signal. What entanglement may allow here is a perfect absolutely unbreakable one time pad. It may allow encryption that is physically impossible to break over any time span (without access to the devices on either end or a side channel attack on those devices that is). It may also allow this without asymmetric key exchange. Symmetric algorithms like AES or ChaCha are still practically unbreakable with a quantum computer as long as 256 bit keys are used, but quantum secure conventionally compostable key agreement is still in the research stages.