Well, D-Wave is actually NOT claiming to have a "generic" quantum computer. Nowadays they are using what is known as adiabatic quantum computing which is much easier to implement than ordinary QC because you only need next nearest neighbor coupling (the quibts are connected in a 2D "lattice") and you do not really need to manipulate individual qubits during the computation (the idea is to initialize the lattice in the right way, and then adiabatically transform the Hamiltonian to the system you are actually interested in).
This is why they can connect so many qubits at once.
The drawback of adiabatic QC is that it can only be used for a few problems, you can't e.g. use it to factorize or sort.
And I can assure you that D-Wave have MANY patents. They insist on patenting just about everything they do which is one (of many) reason why most people in academia (including me, and I am speaking from experience) do not want to work with them (which is a shame really, because they have some really talented people working for them).
The "sealed box" is as far as I know a dilution refrigerator operated at about 50 mK immersed in a liquid He cryostat. So there is a very good reason why you can't open it
However, the actual QC is just Nb structures (Josephson junctions, resistors, capacitors +transmission lines for biasing) on a chip. As far as I remember it was fabricated by JPL using their standard RSFQ fab process. There are some micrographs in their papers.
Anyway, the point is that at there is -at least at the moment- no real reason to believe that they are actually lying about anything. The big question is instead whether or not they are doing anything really
useful, adibatic QC har always been a somewhat controversial idea and many people claim that adibatic QC will never have any real
practical benefits over conventional computers (adibatic QC might turn out to be too slow).
