This seems more like a solution in search of a problem. This is going to add a lot of costs. Let's break it down:
1. New, bigger, more complex connector: adds cost and weight.
2. Bigger carrier board for memory modules: adds cost and weight.
3. Needs screws where before they used latches: Some material cost added and some weight, but worse: a bunch of assembly cost added.
4. Overall heavier than SO-DIMMS, so that weight increase will flow thru the supply chain, incrementally raising prices.
5. New connectors will have higher defect rates and will throw away decades of high-scale manufacturing cost reduction that was put into SO-DIMM.
If there is any possible way to add more lanes or better lanes to a SO-DIMM or similar connector format, especially if it is backward compatible, that would be a huge win over this big, expensive thing.
I don't buy the "thinner" argument. If you want thin, solder the chips on. Everyone who wants thin pretty much accepts that they won't get memory upgrades.