The reasons why Blender isn’t used by big studios are numerous, but they’re simple. Lots of Blender is half-finished. There are tools all over the place in Blender’s trunk that would be considered beta or even alpha state if they were a built-in tool or an add-on tool in another package. That leads to add-on support as well. The plugin API for Blender is quite limited by modern standards. There are lots of python bindings, which is great for little scripts or stringing together existing ops in novel ways (bSurfaces, Rigify, etc.) but it’s not very extensible, and the complete lack of a C/C++ API is a big deal for many companies who need the ability to create fast, reliable tools. Combined with the GPL it’s also created an absolute dearth of off-the-shelf solutions that are common elsewhere. Things like PhoenixFD, Krakatoa, ExoCortex tools, and many others either can’t exist in Blender, or would be completely financially unviable to develop in the first place. The FOSS community likes to pretend like it’s not the case sometimes, but money IS a motivating factor, and you can bet that 99% of the time if a coder has the skill to develop a tool, he’s going to do it where it he can recoup his time in the form of payment.
Finally, this biggest reason, and this has been touched on already: support. Blender has absolutely no official support network in place. And asking them to have one is next to impossible considering the shoestring workforce and budget they operate off of. Production studios want and need this support structure, though. If you’re on a tight deadline and you’ve got important dailies coming up with investors, directors, and decision makers, but you’ve hit a problem a week out that requires immediate attention, that’s a big deal. You need a dedicated support specialist on that case immediately. You definitely don’t want to call up the powers that be and hear “we’re in a bug fixing only phase right now, try the forums/IRC”.
This may sound overly critical of Blender, but I only criticize because I love it. There is potential in the project that’s far beyond what’s possible with the red tape of commercial undertakings. And some tools in Blender are simply the best in class, IMO. I wouldn’t model anywhere else, for example. But there are also lots of glaring errors that will jump out at people once they move past the hobbyist phase of their 3D career. Some of them are fixable, as is the case for the half-finished tools sitting around the code base and the lack of an extensible C/C++ plugin API. It’s simply a matter of available talent, time, and money. Others aren’t, such as the difficulty with getting developers to get in bed with the GPL license. Some coders only work on Blender because it’s GPL. I’d wager that many more wouldn’t work on it because of the GPL’s viral nature and their desire to make money. I know many people don’t want Blender to become a tool where addons add functionality at a cost, but it doesn’t change the fact that money is a motivating factor.
Money, in fact, is what it really comes down to. For a production house, time is money, and whether it’s time lost from lack of support, lack of tools, lack of options, or any number of reasons, most big players have crunched the numbers and found that the cost of licensing an established player in the 3D field is lower than the hidden costs of moving to a pure Blender pipeline. Take that as you will, but it’s unlikely to change any time soon, and I personally don’t think it should. Blender has its niche, and aside from a few niggles it serves its niche incredibly well. I believe there should be a greater focus by the foundation on what they’d like Blender to be rather than trying to hit every single CG bullet point out there, but given the limitations pressed upon it, I think Blender is an excellent piece of software, and I trust the Blender Foundation to keep moving it forward in a positive direction, regardless of whether or not there are suddenly videos talking about how vital it is in a pipeline on YouTube 
