There's nothing implausible about a portable linux machine with those specs and the price doesn't seem plainly impossible. However, manufacturing hardware is very hard and I'm dubious any time a small first-time company claims they've got the situation on lock. I wouldn't give money to anyone who claims they'll give me hardware a year from now because of the massive number of factors beyond someone's control. Even if a company said "We know this is hard, and we've built lots of flexibility into our process so we can still make this if we lose suppliers", etc etc I'd still assume they're underestimating the complexity. I mostly treat hardware projects of any kind as bunk until they're shipping. Not clear to me if the company is trying to make money on the hardware or just exist long enough and get enough buzz to get bought by a large firm.
I think the concept is OK. Interesting that the minimum hardware reqs for low-end PC games have been stagnant for so long that portable form-factor tech have caught up to it. Doesn't surprise me, given that a lot of mobile SOCs are now theoretically in the ballpark of the 360/PS3 (caveat: efficiency losses due to thick, slow mobile APIs versus faster, closer-to-metal APIs on consoles).
Like most specialty portable hardware I can't see it doing huge numbers. This is clearly a company that wants to sell a few tens of thousands in a best-case scenario rather than a few million. The Shield is around 100k in North America based on the recall numbers announced this week, and nVidia is clearly a company with a lot more experience on hardware, better ability to tap supply chains, more marketing (hell, it's in their fucking video card driver control panel). So if you're making something like this that's your "ceiling", roughly
