AGI being "just an engineering challenge" implies that it is conceptually solved, and we need only figure out how to build it economically.

It most definitely is not.