Relationship AIs are at most 5 years away from being good enough for the average man to completely disengage from women. Every day, there is a new AI girlfriend app that has some brand new feature like video generation or voice calling. Instagram has known AI egirls that are constantly gaining followers.
The robotics portion is also coming along, albeit not as publicly. Here is a good example:
We have full motion robotic bodies. What we need are improvements to make the movements more fluid and natural, along with an integrated battery so external cords are not needed. We also need as some point for there to be the development of some kind of openSexBot protocol, so software based relationship AIs can interface with the robot. We should have something like this by 2035 at the latest, and honestly far sooner if we, as a community (meaning AI girlfriend users, blackpillers, mgtows, incels, etc) create incentives like grants and guidelines.
Cost is going to be another factor. I expect the cheapest bots to start at $30k, with the best ones topping at around $200k for the first release versions. You'll be able to get a loan on them, or some countries could get creative and offer them as incentives for populating and working in sparsely populated areas or serving as a foreign legion (Russia).
I've always felt like what we need is a giant yearly conference for the guys in our space and adjacent spaces to meet with industry about our needs. We should raise a 20 million dollar bounty on who can create the first realistic sexbot that can do basic tasks along with performing sexual needs.