Minimal cost?
You have to develop the software, in camera. You have to QA it. You have to update manuals and marketing materials. You have to QA it every time you release new firmware, so it's a gift that keeps giving. There's a good probability that you'll get involved in at least one goat rodeo requiring emergency firmware updates, because you get the DNG wrong in some important corner case that your regression testing missed.
Proprietary RAW means that you can fix a lot of problems in the downloadable free native software, rather than with a firmware update. They mean you don't have to do much in the camera, in fact. That's why they're different from model-to-model within the same lineup, they're trying to push as much work as possible offboard, for excellent reasons.
We're talking 6 figures, USD, easily, to convert from a proprietary RAW to DNG as your standard supported RAW, over a five year period, say.
And what do you get for this? How many sales is this specific feature going to drive? If it costs $100,000 to build and support this for 5 years, which is a lowball estimate, you better show me, I dunno, a thousand units shifted specifically because of this.
And you and I both know it's not actually gonna shift a single unit. Sure, put it in, and Michael will sing your praises, but he still won't buy the camera because you don't allow him to microadjust the temporal thrusters or something. And every guy like Michael will have their own reasons for not buying your camera, because ultimately, the care about stuff like lenses and weight and shooting speed and color renderings and resolution and and and and not about minor convenience features.
The best you could hope for is that three guys who were really torn between camera X and camera Y will finally tip over to yours because of DNG. If only you'd spent the $100,000 reengineering the body to pull out 17 grams more weight, you'd have sold 1000 more units because you'd be lighter than camera Y instead of the same weight. Ooops.