At this rate the only way CoinTerra is going to be worth is if they produce a 1TH/s machine and price it so good that nobody will even think of KnC. It has to be the same sort of difference like when you see BFLs 500GH/s machine sold for $22k when KnC is offer 400GH/s for $7k - it's a no brainer. Cointerra is going to have to pull off something like this. They openly stated that they want to make it more efficient with smaller chips etc, but with Bitcoin and where the difficulty is headed, that won't matter. Right now being first counts, because at the rate the difficulty is set, by the time the CoinTerra machines arrive, they are just going to be too late and overpriced.
To be honest, ASIC manufacturers should really be talking about 750GH~1TH/s at this point in time.
It is so funny that everybody out there (KNC and now CoinTerra too) tries to put as much hashing power as possible in a single package, like a race to say "look, I got the biggest one!". But you know what? At the end it doesn't matter if your ASIC does 10, 100 or 1000 GH/s, the only thing that matters is the total cost of the BOM, including external components.
As such, a $50 power sucking 500GH/s ASIC can easily be beated by an array of 10 highly power efficient 50 GH/s tiny ASICs that cost $3/ea. Ah, the magics of distributed work!