I think this is a big deal. A really big deal.
Although I've only dabbled with Arnold, it's clear they mean business, it's a first class raytracer, with a solid portfolio already under its belt. What makes this so big is that this positions Arnold in such a way where it could well become the default renderer for many artists seeking to take their work to their next level.
Although mental ray in its own right is a capable tool, it's no VRay. I have always felt the mental ray is something of a red herring for beginners when it comes to learning advanced rendering. It's there, so people use it, but it's by no means the best tool on the market. Not by a long shot. Although I actually always really liked using mental ray, it always felt more aimed at programmers than artists.
Now Arnold has the kind of capacity where it can rub shoulders with VRay and the proprietary renderers big studios develop. If its market penetration is significant enough, it could become the new standard which drags the industry a whole new direction.
The big barrier to artists trying new products isn't so much the potential quality of the new product, it's simply offsetting any potential benefits the new product might offer against how long it may take to master it. Corona renderer is a classic example of this. It's a beautiful raytracer, it's simple, elegant, easy to learn and fast (relatively speaking). It's also priced significantly keener than VRay. Despite all this, I've only ever used it for personal projects. Because I'm just not going to make the leap to a new tool on live projects unless there's a very clear and strong benefit.
What if people develop this mentally in favour of Arnold?
Needless to say, if this results in 3DS Max integration I'll be one of the first to take it for a spin in the arch viz world, having already enjoyed playing with it in Maya. I'll be also very interested to see what effect this has on things over at Chaos Group...