Yes, that's exactly what I'm saying. If they GPL ZFS, they end up risking a major reason people choose Solaris. ZFS is very much a "brand" among systems administrators. It took quite a while for Sun to convince people to put their precious business critical data on ZFS, and it'll be the same story for btrfs.
Both of these projects are costing them so little, it's much easier to justify keeping them going than it is to take on the unknown business risk of making ZFS directly merge-able into the Linux kernel. Sun clearly chose the CDDL to keep this from happening, and post-merger, you've probably got the same people in charge of the Solaris and ZFS projects, so strategies aren't going to change overnight.
tl;dr: Oracle has low, predictable expenses for development of both btrfs and ZFS, and there's no reason to rock the boat for <1m/year in expenses.
Both of these projects are costing them so little, it's much easier to justify keeping them going than it is to take on the unknown business risk of making ZFS directly merge-able into the Linux kernel. Sun clearly chose the CDDL to keep this from happening, and post-merger, you've probably got the same people in charge of the Solaris and ZFS projects, so strategies aren't going to change overnight.
tl;dr: Oracle has low, predictable expenses for development of both btrfs and ZFS, and there's no reason to rock the boat for <1m/year in expenses.