The system operates at 20 milli-Kelvin (0.02 degrees above absolute zero), which is 100 times colder than intergalactic space. In fact, his machine is probably one of the coldest points in the universe, unless there are any other lifeforms out there building similar machines. The machine is incredibly well-shielded from stray magnetic fields so that only one other device on the planet has a better ‘magnetic vaccuum’. He was practical in his descriptions, showing us circuit diagrams, explaining the physics, describing the sorts of computations we might be able to do on this device.
Basically, my take-away message was that we will be able to do computations involving many many more parameters than previously possible. Even now, he’s promised the next upgrade of the chip (going from 128-QuBits to 512-QuBits) might allow us to speed up a computation that would take 320,000 years to perform classically to a mere 120 ms. Naturally, this was talk for venture capitalists, but still, awesome stuff. The way we need to frame our thinking is to use the tools of Machine Learning approaches that many in the AI community use already. OK, this, I think, is something we can do. Already, colleagues at USC have worked on preliminary studies on ‘Quantum Adiabtatic Machine Learning’ on this system, and it was a sobering moment when Geordie looked at us all with a serious expression at the end of his talk and said: “What I want from you are Nature and Science papers that use this machine and demonstrate its capabilities”