It stops being low-margin if suddenly you can build things much cheaper than your competitors can, and they can't replicate what you're doing.
Maybe the insight is that it is better to first sell the robots (the "shovels") and only later go into the gold business yourself (if ever).