Looks like a great idea, with a few potential problems:
1, the footprint is still massive compared to a TextBlade, and there's no way to make it fold.
2, they have to put in 80 keytop touch sensors, instead of just 8 plus however many are in the spaceblade of a TextBlade.
3, I can imagine a few cases with that where the keyboard confuses mouse and keyboard input, i.e., you lift your hands off the keyboard for a bit to do something else, reach out to start typing again, and fling your cursor somewhere completely unexpected with an errant mouseclick on the keyboard, or are mousing around and push down a little too hard--hiccup, maybe?--and toss out a random character or two. There would have to be some way for the system to tell those different intents and shaking hands and suchforth apart, which adds further complexity and Murphy bait to the signal processing for those 80 touch sensors.
4, I'm not sure how the gaps between keys might mess things up either in terms of mouse response or how it feels to the user, but it does seem like a potential problem, and this from someone who actually likes a little texture to my touchpads.
5, I'm guessing they have to have very very flat keytops, possible not even the usual bumps on T and N (F and J for anyone still using Qwerty instead of Colemak). Flat, non sculpted keytops don't feel as good and don't give the user as much feedback about where their hands are on the keyboard. If they put the keys really close together to counter problem 4, it worsens the symptoms of problem 5.
Hopefully they did a good round of testing to iron that all out, and I wish that keyboard all the best at getting to market in a timely fashion and turning out a decent product as well.
Makes me think of another similar thing I saw a while ago, which rather than having 80 individual touch sensors, just used a grid of invisible lasers at the edges of the frame to turn the surface into a touch panel. I have no idea what that was called or how successful it's been since I saw it though.