As someone who works in house for a SaaS company, we have been migrating away from Bootstrap and to our own framework that we have built on top of Bourbon, which is sort of a framework for building your own framework.
We did this for two reasons, primarily.
First, we didn't want to carry the weight of the entire framework - both CSS and JS - on every page. Yes, I know you can pick and choose, but for us, it was easier to build up to what we needed rather than eliminate down.
And even if we were to cherry picked, we would still be left with bloat. Take the grid, for example. We only need maybe four or five layouts, almost all of them based around breaking the page in to thirds. Why should I carry the code for all the other column layouts if I only need a couple?
This leads to the second reason, which is rolling our own allowed us to customize to our needs specifically, as opposed to either living what we were given or having to hack around it.
For instance, our modals look the way we wanted them to look, using the class names that make sense to us. Bootstrap in particular can be a bit overly tag heavy in my opinion. And overriding something like a modal can be a significant investment - one that has the chance to break every time you upgrade Bootstrap.
So we roll our own, using the BEM naming structure and Bourbon (and its related libraries) as our helpers. Does this mean we have to do a bit more hand holding with people who are not as experienced in the front end area? Absolutely. But that is a cost we are willing to pay to ensure that the front end code adheres to the guidelines we want to follow.
All that being said, Bootstrap and Foundation are great if you want to just throw something together quickly. Rolling your own takes time and a solid understanding of where the overall design is headed. For prototyping or even getting the first version of something out quickly, you really cannot beat a framework. The benefits in those situations easily outweigh the costs.
But if you have the team and the right situation (long term maintenance of an app), I suggest you consider taking the time to do it yourself.