The part that you're missing is that something can only be a "derivative work" if it itself is copyrightable. There is a general understanding that software binaries are copyrightable, and this has obviously been tested in court. ASIC netlists are not the same. *If* they are copyrightable *then* yes, as a derivative work of the source, they'd be copyright the author of the source. The question is whether they're copyrightable or not.
You can't assert that something is copyrightable because it is a derivative work, that's backwards. You have to establish that it's copyrightable first, then you can say it's a derivative work.