you can incrementally add documents yea
index = VectorStoreIndex([], storage_context=storage_context)
for doc in documents:
index.insert(doc)
However, you can probably speed up the emebddings step. What embedding model are you using? You can increase the batch size