Constant returns to scale are mostly mathematical convenience. Empirically they can occur but there is no guarantee that they will. Although you should note that Solow model can be tweaked to accommodate both increasing and decreasing returns to scale. See Neto, Claeyssen, & Júnior (2019) paper as an example of people doing that, so this assumption is mainly there to make the model easier for students.
For example, Basu & Fernald (1997) found in their study that on average US firms in their sample had slightly decreasing returns to scale (although they are quite close to be being constant). However, this is by no means something that has to hold it is just empirical observation. Empirically you can have constant, increasing or decreasing returns to scale and it depends on the time and place.
If you want to read more studies than just the ones I listed above you can simply use google scholar using the keyword returns to scale and then country and period you are looking for (if you would want to know it on more micro level you can add industry). The literature on this subject is too wide to provide any exhaustive overview.