Having to pay expensive diy-developers is also a cost (and a huge risk). And if you then need to transfer knowledge and reinvent the wheel for all the NIH-appli

Having to pay expensive diy-developers is also a cost (and a huge risk). And if you then need to transfer knowledge and reinvent the wheel for all the NIH-applications you're essentially becoming a cloud, but less complete and not scalable at all.