Absolutely. People on X keep making the mistake of assuming cloud / network latency is the problem here.

The vast majority of America is within 10ms of a data center. That's nothing.

The current challenge for most interaction is ASR -> prompt processing latency. This will be improved with multimodal models on specialized hardware like Groq.