Inside Microsoft's 'soup to nuts' quantum computing ramp-up
- 28 April, 2017 06:30
Professor David Reilly
The concept of building a quantum computer with topological qubits had been knocking around Microsoft for nearly 15 years.
At the company’s Redmond campus, later expanding into its Station Q centre at the University of California, researchers investigated this “complex, yet beautiful mathematical theory” as Microsoft’s quantum guru Michael Freedman once described it.
It remained just that – a beautiful theory – for years. Then something changed.
Suddenly in late 2016, Microsoft announced Station Q would expand to eight labs worldwide including one at the Quantum Nanoscience Laboratory at the University of Sydney. Four heavyweight academic hires were made. An 'inflection point' had been reached, a release stated.
“Why are they now switching on? How much should I read into that?” says Professor David Reilly, director of Station Q’s new Sydney lab. “I think you can read plenty into that.”
Is Anyon out there?
The topological approach to forming qubits uses quasiparticles called non-abellian anyons.
“We've realised that the only way to really scale is to build much better qubits. Qubits that are inherently, intrinsically immune to noise and immune to disturbances from the environment,” says Reilly.
In most approaches to building quantum systems, information is encoded in the properties of particles. That makes the systems very fragile as any disturbances from the environment can destroy a particle’s quantum state.
According to Microsoft, topological qubits are better able to withstand heat and electrical noise, which allows them to remain in a stable quantum state for longer. By encoding the information not on the quasiparticle but in the order the positions of the anyons are swapped around (called braiding), topological qubits offer a more viable way to make a scalable, usable quantum computer.
The quasiparticles themselves remain something of a riddle. Non-abellian anyons can’t be seen with any kind of microscope but they can be measured with high-precision devices. Probably. Certainly there is no consensus among physicists as to whether they are real or not.
As Microsoft’s Alex Bocharov confidently put it to Nature, the company is “pretty sure” that they exist.
“We reached a point where the experiments were convincing enough that now we should actually try and construct technology from it. And that's where we are today,” Reilly adds.
Faith at scale
Backing topological qubits is as much about belief in the approach, Reilly says, as doubt in all the others.
“It's a double-edged sword. For me personally some of the decision to go in this direction is born out of a pessimistic view. It's faith in this approach and the reality is it's pessimism that the other existing approaches can really go the distance,” he says.
Google and IBM back the superconducting loop approach to building qubits, while Intel and the Commonwealth Bank and Telstra backed Centre for Quantum Computation and Communication Technology at UNSW in Sydney, are pursuing a silicon-based method.
“The difference between demonstrating the basic operation of one single device in a university lab and publishing a very nice paper, and what it's going to take to bring that technology to scale – I think these are vastly different activities,” Reilly says. “It becomes pretty depressing and I don't see it.”
The ability to scale is key to Microsoft’s quantum efforts. Its ambition goes way beyond the lab.
“It's all about having the fundamental building blocks that can take you to millions of qubits,” Reilly says. “It's all about going the distance.”
Page Break
Mathemechanics
A lifelong academic, Reilly worked as a consultant to Microsoft for a couple of years before joining them full-time in October. Accepting the offer – a job Reilly calls “a perfect mixture of motor mechanic and mathematician in one” – was easy.
“You're dealing with a big company with a lot of smart people, a lot of resources, and a lot of experience in bringing technology from a theoretical idea to a reality,” he said.
“We are now thinking about quantum computing not as an esoteric research topic that's a curiosity but we're thinking about it from the point of view of actually building, constructing a machine.”
That machine can’t run on qubits alone, however. Reilly and his team are working on the interconnects that will pair qubits to classical systems. It’s no easy task, and requires a back to basics approach to classical computing. The solution will also need to work in the near absolute zero environments quantum systems operate in.
“It's back to 1936, before ENIAC and before the large-scale vacuum processors before transistors, microprocessors... We have to go back to some basic understanding of what computing is,” Reilly says.
Soup to nuts certain
Station Q in Sydney recently announced it was going to double its headcount by taking on 20 or more engineers. The 'doubling down' on quantum, as the company puts it, is taking place across the business.
Long-serving Microsoft executive Todd Holmdahl – who was involved in the development of Kinect, HoloLens, and Xbox – has been put in charge of the scientific and engineering push.
“A complete spectrum” of expertise has been hired worldwide, Reilly says, “from soup to nuts” (a phrase he picked up from his new American colleagues).
“The reality is…you are not going to build a quantum computer in your backyard and you're not going to build a quantum computer in a university lab with your graduate students and the kid that took a third-year project with you.
“It's something that requires a very large number of people, deep connections and networks throughout the world in the technology space, existing frameworks and contracts for getting things built.”
In other words: Microsoft has what it takes to build a quantum computer. But when?
Not soon. The challenge is harder than sending humans to the nearest star system, Reilly says. It is “the hardest thing we know of”. But it’s not impossible, and within our lifetimes.
“I don't want to retire with it still living in a world of academia and ideas, even if they're very interesting ideas…for me the dream is that we see this stuff actually come to life,” he says.
“We have turned a corner in our understanding, in our thinking and in our ambition to construct the technology and we think that now is the time to move on it.”