The problem with anything SpaceX/Boeing/private tard company is the absolute lack of anything resembling quality assurance or systems engineering. They are for profit and don't see a problem with a 3% chance shit explodes because who cares about testing the raw materials for purity when skipping that means saving $50k (despite having three rockets explode as a result costs $1.5m). This design model goes tits up when people get involved
The main problem is basic mathematics. You don't actually gain that much when the satellites need to be specialized for a certain roles as they can cost billions. And when you are spending idk 3 billion on a payload there's really little difference between 120mil for what would have been an Atlas V or 60mil for a Falcon 9 (my launch costs may have been off but that's the rough ball park). You can get certain capabilities out of Falcon Heavy, sure we have Europa clipper. Launch vehicles matter in terms of capability, less in cost.
This is why the design method fails, because it assumes the market will grow with cheaper launch vehicles which is an "ehh" because all the hardware for payloads needs to also become cheaper.
Yes, for refuelling or something like Starlink (or even dumb big satellites) you would want something like Starship - if it ends up working as advertised, which is a big if. The Starship has been a farce so far in terms of development and has shown very limited orbital capability. Future versions are meant to improve on this, but it is a big if. Now, this is not to say refuelling won't do a lot. It will, but it is all something that needs to be worked out.
Neither did the space shuttle. Did Orion? Does this thing? Does Boeing's alternative?
The problem with the Shuttle is that it was forced by government pressure into being what it was. But, it did have failure modes. Most them absolutely horrible, but they did exist. The crew of Challenger are believed are believed to have survive the initial catastrophe. The reason they couldn't have survived? No parachutes, no pressure suits and blue flightsuits. No parachutes is obvious, same with pressure given the whole oxygen thing and well, you don't want to blue out in the ocean. Hence the orange shuttle suits.
There are also other shuttle modes that have been performed, Abort to Orbit has been done before. And there were concepts such as Return to Launch Site. The Shuttle can also glide and reduce speed at least allowing a crew to bail out. Albeit, most of the came about because of the SRBs and how you had to basically stick with them. Which, could bring about structural flaws if the Shuttle's main engines all failed. But, there were even abort modes for that. Case in point:
The Split S abort.
Which yeah, these abort modes are all fucking insane. But at least they could exist, and maybe a chance would exist. This really is not the case on Starship which is reliant on engines out.
And now onto Orion, Orion has plenty of Abort modes. It is a very traditional, safe capsule. This flight is generally one of the safest flight paths possible.
First abort mode, holding the flight. Not going through with it, the second abort mode:
The third abort mode, the launch escape tower which would fling the craft away. It has been tested multiple times and has been confirmed to work. The fourth is an abort to orbit, which would use either the interim cryogenic upper stage or the service module. The fifth would occur following the ICUS allowing Orion to return with ease. Then, when Orion does it's trans lunar injection it would be on a free return trajectory meaning it will return to Earth without any other burns (but Orion will perform minor corrections on the return leg).
God Bless