I think I didn't detail that aspect. In most laptops, when a dGPU is used, the gpu chip itself and memory are all soldered to the mobo and connects through an internal PCI-e x16 interface. This occurs without a slot. What I'm proposing is doing the same thing as above, but with an SSD on the mobo itself instead of the dGPU.
This should have the effect of reducing cost for SSD use, increase performance, and keeping size down while still having both an SSD and HDD in the same chassis.
An SSD and HDD in the same chassis already exists in every conceivable definition, and isn't too hard to get in really tiny form factors, if you really want it. But, most people don't. I have both in my mini tower, but given my storage growth, if a 2TB SSD got to at or under $300 any time in the next 5 years or so, I'd get rid of it, putting even "slow" data on the SSD. If going SFF or mobile, I'd use an external drive or network storage. If stationary SFF, networked drives would be sufficient, with GbE. External HDDs aren't too reliable (even assembling them yourself), but they beat internal notebook HDDs, IME (and for tablets, an HDD is a non-starter).
Cost-sensitive devices already use soldered-on flash storage. It's usually eMMC, but if cost allows, PCIe would work. UFS is already in the wings to replace eMMC, too. Even real Windows runs on eMMC (albeit running like shit, due to poor write speeds).
If it's expensive enough not to use low cost embedded flash, though, it's expensive enough to allow storage customization, replacement, and upgrading. Not everyone wants a Chromebook, and only Apple could charge high prices for crap like that and get away with it. Notebooks with mSATA or M.2 SSDs and 2.5" HDDs, using the SSD for caching, have been available from the major OEMs ever since mSATA came out. Before that, there were dual-2.5" notebooks with cache drives. That is preferred over soldering it in.
It's not a technical limitation keeping OEMs from integrating SATA or PCIe SSDs, but that there is a threat of market forces going against them if they do, since people expect to be able to swap those parts out (or, usually, have their computer guy do it), and that it could add overall costs that may be higher than the cost of adding a slot and buying the parts. It also allows the OEM to have choices in terms of what storage device they use from whom. Finally, it allows them to not have to R&D that part of the computer, which is not yet mature enough to do easily, for high performance SSDs like we have in our PCs.