Mars bases and colonization is what Musk tries to sell. Similarly asteroid mining, data centers and factories in space, and massively increased rollout of satellites. All of these are stupid because they all rely on massively over-inflating the upsides and really thoroughly underestimating the costs and expenses of doing things in space, even if we could theoretically put some of these things up there. The only plausible part of the plan that makes sense is increased demand for satellites, but that includes a lot of double-dealing, since Starlink is going to be a supposed driver of that.

and everyone getting their internet from starlink

Similarly I don't think it is theoretically impossible. I have not looked at the numbers or how much of the electromagnetic spectrum that that would require with current technology, nor have I seen anyone else try to address it. But my feeling is that this is unlikely to work at that scale and would be prohibitively expensive to keep such a large covering of satellites in low earth orbit, aside from other issues like making ground based astronomy impossible and risk of space debris. I think this is the same thing as everything else, of just really trying to low ball and ignore the actual costs and just rely on the idea of "Musk launches a few more satellites and everyone can just get internet everywhere and we don't have to maintain all of this pesky ground based infrastructure like wires and fiber"