"There are lots of folks around the world doing research projects at this level, where they run for a few minutes and then write up their results, but we've had to run 24x7 for years at a time," Brownell said. "Lockheed Martin, our first customer, came on in 2010. There are a lot of challenges in combining ultralow temperatures with enterprise quality levels."
Besides just quantum computing hardware, D-Wave also works on software that takes advantage of it. For instance, it's developed software that injects quantum-computing capabilities into a machine-learning training algorithm for faster training times and better accuracy.
"Adding quantum computing into your classical workloads will provide an advantage," Brownell said, citing other examples including portfolio analysis, pattern recognition and optimization.
In general, when a user models a problem using D-Wave's technology, the processor considers all possibilities simultaneously. Multiple solutions are returned to the user, scaled to show optimal answers.
D-Wave has just a handful of reference applications that can show customers how a particular task can be accomplished, but it hopes to expand that number significantly.
"People shouldn't have to understand physics at all to use these tools," Brownell said.
D-Wave and IBM have gone head-to-head in the quantum space in recent years, and there's been considerable debate over whether the technologies actually live up to their claims. Further complicating things, the companies take very different approaches, making it difficult to compare them.
With a focus on bringing a product to market as quickly as possible, D-Wave opted early on for a model focused on what's known as quantum annealing, in which the technology uses quantum fluctuations to solve a particular type of problem. IBM uses what's known as a "gate" or "circuit" model, Brownell said.
That model is "reasonably elegant and makes a lot of sense," he said. It could also be more broadly applicable.
"The gotcha is that it's super hard to do," he said. "I admire the research by IBM and others, but it's going to take at least a decade before there's a product that does anything useful."
For example, the technology IBM recently offered up for public consumption features five qubits.
"Will anyone ever be able to build a gate model with 10,000 qubits? That's an open question," Brownell said. "When and if that model becomes implementable, we'll have the building blocks in place and will have tackled the hard problems before anyone else."
Yet another approach is known as the topological model of quantum computing, and that's the one Microsoft has taken, he said.
"It's actually more elegant from a theory point of view, but it will require the discovery of new kind of particle that no physicist has ever seen before," he said.
Sign up for CIO Asia eNewsletters.