JP: How is your commercial approach different from the academy’s?
GR: The academic’s approach is not necessarily worse than ours, but it’s different. Our approach is to throw as many qubits as you possibly can onto a chip, get it solving real problems, and then use the performance on those problems as the metric with which you gauge what’s better and what’s worse. So when you increase the capability of the machine, you are increasing the capability of the machine to solve problems faster, and bigger problems. Compared with the academic approaches, ours is quick and dirty, although I don’t think it’s any less careful.
JP: What kinds of things could I do with a 1,024-qubit quantum computer?
GR: There are lots and lots of existing commercial applications that require an optimal solution to a problem with a lot of variables. For example, in chip design, a lot of the issues that have to do with hardware design verification are of this sort. There are also lots of applications in financial engineering that investment banks have been very interested in pursuing with us: things like portfolio optimization, risk reduction, selecting and pricing derivatives. In addition, every single scheduling problem that exists in the world is one of these problems. You can imagine somebody like an airline or a federal-government organization that had to schedule lots of people where there are all sorts of issues about who works where and who gets access to what and why. These problems create these massive conflict-resolution scenarios that simply can’t be managed nowadays. They’re too hard to solve in the lengths of time that people want to solve them in. I think that the way it’s going to look in the future is that anybody who has a significant scheduling, routing, planning, application–all of those applications will be ported to our machines, which will be available online.