D-Wave just recently "demonstrated" their 16 qubit prototype at the Computer History Museum. It has understandably garnered quite a lot of press, particularly amongst the geekier elements of the World Wide Wibble. From what I've read, they are trying to build an adiabatic quantum computer from superconducting qubits (Josephson junctions). Other quantum information scientists have commented on the plausibility of the device actually working, some more optimistic than others. I would class the general plausibility as just above Steorn, but I will reserve final judgement for if or when they actually release technical details, like temperature, energy gaps, decoherence rates, speed of operation, charge fluctuator noise levels and algorithms run. If they do actually manage to build a 1024 qubit device which is controllable by 2008 (and I don't mean just a bunch of SQUIDs connected to each other), then I will be very, very, very surprised.
I hope that the whole media circus surrounding the company and its claims will not seriously damage quantum information and computation in general. Inevitably, when commercialism meets science, science usually takes a back seat to publicity and PR.