Sony owns the market for mobile/dSLR sensors, RED make their own with TowerJazz (and their R&D cost must have been monumental, hence the high prices) and blackmagic uses CMOSIS sensors, which I'm sure is half the reason they can price the cameras the way they do
edit: sensor in the BM Ursa Mini is based on this, not CMOSIS anymore - http://fairchildimaging.com/catalog/focal-plane-arrays/scmos...
Each has weaknesses. For example, Red's build quality isn't all that great and their lack of awareness for what cameramen need on set is lacking. ARRI has a price issue, as well as resolution and raw left to be desired. Especially their AMIRA camera, which could be a first target for a startup - a good documentary camera. Blackmagic is usually all talk - their software is abysmal and hardware has a cheap taste to it. Sony lives in their own world (their professional market division is a bit out of touch with reality).
In my opinion, there's a great space to be filled (at first, but there are others) in documentary and/or ENG camera space. Making a good ENG camera paired with a lightweight version of codex at a reasonable price would be a killer combo. Two main (of many) issues are CMOS design and production and either custom lens manufacturing or sourcing those.
I am not even sure where one would begin to implement their own CMOS. Everything silicon production seems like alien, otherworldly, tech to me once you start thinking of production. Design is on another area51 level as well.
Back to original issue. Resolve and Fusion are not originally BM code. Resolve has got some issues since it got acquired (stability for one). BM cinema camera overheats and is barely usable in a production environment, pocket camera is a nice toy but lacks software features (for example - take a one picture only). I used both and will not use them again. I will have a look at URSA once it will be available in general. I was very interested in their scanner as well, but I am starting to think it's vapourware at this point.
What people outside don't understand is that camera is only a part of an equation. That's why you see lots of amateurs buying their own Scarlet, Epic, whatevers and thinking that's what they need to shoot a film. I'd rather have a slightly worse camera and good lens than vice versa. Then there's grip, light, people that know how to use all of that and have experience...
Regarding cameras, there's still a lot of space to improve on and if one could either source good CMOS sensors or make their own - this could easily be a space for a startup to thrive at.
edit: What ARRI did right, apart from tonal range (which isn't an issue with Red and a good colorist) is that they have tons of experience how to do the hardware right that plays well with grip.