I think that's the opposite of what the GP is saying. The thesis seems to be that hardware is patentable because what matters is not the schematic/topology, but the physically manufactured circuit board, whereas in software you can directly copy an algorithm. It's the "you wouldn't download a car" differentiation.