The rule of thumb for most complex traits is half genetic, half environment... but this half is almost all "un-shared environment", meaning things specific to the individual (possibly random), not "shared environment" meaning observable things like parents, diet, schools, typically shared between siblings.
I don't know the numbers for autism, but would guess it's like this. Which is to say, be careful how you read statements that nurture plays a role -- it's often written to mean "not-nature", and but read to mean "upbringing".
Genes can be activated by the environment, as early as the womb. So in the case of the twins, a brief shortage of nutrients or oversupply of nutrients, may cause certain genes to not active, active late, activate differently, etc.
From what I gathered on the studies around it all, most of the outset is defined by genes, the environment then can play on the existing genes. Ie, if you don't have the genes to be smart, you won't be smart and there isn't much you can do about it past a certain point, if you do have them, you can be smart or you can still be dumb without being able to do anything about it. Or height. If you have genes to be 2m tall it won't help you much if a nutrient deficiency in early childhood stunts growth and leaves you at 1.6m.
So in the case of the identical twins, they likely have both a potential to develop autism or at least tendencies towards the spectrum but only in one the specific genes activated for whatever reason.
So those of us who strive to do better than the lizard brain should be taking up the nuanced, moderate view.