When you see a tech demo like this, you can generally assume that this is what the next generation of fighting games will at least approach in terms of fidelity and realism. Demos are a tricky thing, because unlike games, you can get away with rendering only the small environment you're currently looking at, and moreover, you can optimize the environment for viewing it only from that limited perspective, making it appear hyper realistic. This is why fighting games tend to look better than just about anything else out there - they're the closest a videogame will ever come to these tech demos in terms of being able to "cheat" like this.
Which games are at the other end of that spectrum? I'd probably have to say MMOs. By necessity, they end up looking a generation behind the latest state-of-the-art for two important reasons: First, naturally, they tend to set the minimum system requirements a bit lower to be more inclusive and attract a bigger customer base. Second, and more importantly, MMOs spend a ridiculous amount of their rendering budget on drawing the large numbers of unique characters on screen at any one time, as well as all the effects that can be fired off by them, and of course, any NPCs in the area, and finally, a typically spawling, open terrain to explore. As such, you can't expect an MMO to look like a AAA single-player game, because the rendering budget is spent in significantly different ways. So, I guess you can expect MMOs to look this good perhaps in another two to three generations.
As a game developer, was I impressed? Well, yeah, as much as I'm impressed by all our modern technology. Nowadays, it's actually pretty easy to sink your entire rendering budget into a single character (or small numbers of characters) and make her look quite impressive - she still looked good though. I was less impressed with the outdoor shots. Short of simple interiors, bare, rocky terrain is the cheapest and easiest type of terrain to build and render with the best looking relative results.
I'd call this a decent, incremental step forward, and I'd say that's a good thing. Radical leaps means everyone has to re-invent their entire production pipelines, and that takes a lot of focus away from where in needs to be, which is first and foremost in creating a fun game. In terms of creating game assets though, the major steps forward that need to be taken are how to build more high-fidelity assets for less - which right now is insanely expensive because it's nearly all hand crafted, and so far have always needed to be re-created entirely from generation to generation.