A type of fusion even less advanced than the fusion that has been 30 years away since 1970? That’s at same level of reality as science fiction writers in the 1930s assuming that everything would be powered by atomics.
Maybe you’re just not old enough to be jaded by all the technologies that were in laboratories, promised everything, and achieved nothing. I’ve seen it for the 50 years I’ve been paying attention.
Energy production is hard. Really, really hard. Nothing game-changing is on the horizon.
The only real thing I’d add at this point is that self-replication isn’t likely, not unless that replicator has access to, or can synthesize the ‘spices’ necessary to duplicate themselves. (They’d need foundrys, chip fabs, deposition chambers, doping, etc…and that’s only if they were as technically complicated as a cellphone.)
Ok. So one way to view this is just a matter of scale. If you could put a single copy of all the modern infrastructure you mentioned into a large, warehouse sized building, and every human hand that would normally involved was just a robotic arm with fingers, would you not agree you’ve made a crude and slow self replicator?
It might be 10 miles x 10 miles or something stupendously huge, and it might take several years to copy itself, but you’ve done it.
So then the obvious thing to do is say, how much of all that equipment that is so bulky now can be replaced with more flexible machinery that is more general purpose. Replace casting systems with just cutting the metal parts you need using multi-axis CNC. Replace plastic molding with high speed 3d printers.
Replace unique industrial processes for every feedstock with plasma separation for all inputs. Stop using robotic ‘hands’, develop more refined ways to do it. Go to 3 dimensions. Without workers needing to be protected from falls, you can have your machinery stacked on top of itself. Remove all the air from the building, it just slows down the equipment and is a source of contaminants. With all the air removed, heat losses are lower, so start replacing electric motors and wiring with faster and more powerful/compact superconducting motors cooled with LN2.
And I’m just talking low hanging fruit. What could you do to make this factory faster and more compact if you had more mental ability than all the human engineers alive in the world would combined?
Now, the logical end result is nanotechnology. That might be thousands of generations of improvements later, but it’s basically the compression of all that complexity down to some machine that is the size of a refrigerator. It’s not any simpler - a refrigerator sized nanofactory would have billions of internal assembly lines, complete with nanoscale equivalents to the macroscopic robots and assembly lines and all the rest.
I only mention it because it’s a known endpoint. We are reasonably confident it could eventually be done, compressing all the complexity of modern industry into something that would be small and hyperefficient (and rather convenient to pack onto a starship…), but it might be possible 10 years after the start of the Singularity or 1000 years. Nobody knows. We also don’t know if you could do better than nanotechnology, but we’re reasonably confident the physical laws of this universe would allow a piece of machinery like this to function. (yes, there’s entropy, but every production path in the machine is duplicated multiple times. Eventually, parts inside will fail, but the machinery would have fabricated at least one additional copy of itself before that happens, as well as whatever products it is ordered to make. So in a way this type of machinery would use a strategy similar to living cells, duplicating itself to get around damage caused by entropy)
This is what I was trying to say; plus, some tech just isn’t really portable, needs to wait for humans to commit to year-round settlement. Perhaps a culture could get higher-quality pottery by building a rock kiln at various locations on their seasonal circuit; maybe even that would double as a smelting operation. But some equipment, like papermaking, more advanced woodwork, etc. require a bevy of tools and raw materials, time, and especially expertise gained from dedicated practice. All that requires a permanent settlement - and that usually requires agriculture, or else population pressure if not resource depletion will mean a move eventually.
The only basically (primarily) non-agricultural society that comes to mind was the northwest coastal Indians of Washington and BC through Alaska. The sea and rivers provided substantial food so they established permanent settlements, built giant wood houses, etc.