Into the Light: Xanadu’s Vision of a Photonic Future

Xanadu’s Nathan Killoran accepts the scale of the company’s vision: in two years, this two-year-old company of around 35 software and hardware engineers is dedicated to building a scalable photonic quantum computer with accompanying software.

“We’re trying to solve monumental technological and scientific challenges,” says Killoran, “It’s quite a big vision; it’s a huge problem to solve.” 

The decision to perform quantum computations using photons rather than electrons is Xanadu’s unique selling point. Although it presents challenges, photonics is also an established field where existing technologies, widely used in telecoms for example, can be leveraged. As Killoran puts it: “The difficult problems we have are different to the problems that other quantum computing companies have.”

One major advantage which Xanadu’s photonic computers would have over electron-based systems is energy efficiency. Where most quantum computers operate in dilution fridges at just above absolute zero, photonic computers can theoretically work happily at room temperature without the risk of decoherence, a huge technological benefit.  

However, like other technologies, there are some physical obstacles, which the Xanadu team must currently code around. Chief among these problems is ‘loss’.

“With photonic quantum computing, you encode information in the quantum states of light," Killoran says. “As you shoot this light down a fibre built onto a chip, there’s some probability that the light will lose amplitude or lose photons. This degrades the quantum computation - and if you have too much loss, it degrades to a point where it’s no longer doing something interesting."

Xanadu does all design, testing, integration and deployment of chips in-house, leaving the non-core manufacturing element to the chip foundries. Currently the company has two software offerings: PennyLane and Strawberry Fields.

“Strawberry Fields is basically the programming language or API to work with our model of quantum computation,” says Killoran. “If you want to work on a computer that’s based on light, you can’t use the software that’s available from our competitors, because those are dedicated to a different model of quantum computing. Basically, Strawberry Fields is the first software that targets this light-based model of quantum computing. It has three different simulators that target different use cases;  it means we can take all of the ideas from machine learning and test them on our simulator.”

There are multiple potential applications for Xanadu; and its software’s focus on machine learning lends itself to some familiar quantum applications, particularly drug design.

“I don’t want to tip our hand too early, because we have some ongoing research projects,” he says. “But think of it like this: drugs are materials. They’re made out of physical building blocks and at a fundamental level they are described by quantum mechanics. So, the strong belief in the field is that if you have a device that is also quantum mechanical, then you can simulate molecules and chemical processes and vibrations - the kinds of things you need to do to understand chemistry at a fundamental level."

Killoran says that Xanadu hopes to have its first   quantum computer running inside 24 months. In that time, the team will be focusing ever more on the technology’s machine learning applications, laying a foundation on which disciplines like drug discovery can then build. But like many companies operating in the quantum space, Xanadu also has a more personal goal: to build the first chip to show quantum supremacy over classical computers, or what Killoran describes as quantum’s “big bang” moment.

Whether it’s Xanadu or one of its competitors, a company showing quantum supremacy is, Killoran says, a reasonable scenario in the next two years. It will also be the catalyst that begins an explosion in quantum computing uptake, with quantum computers taking on increasing varieties and amounts of grunt-work behind the scenes, separated from commercial users by apps and interfaces in the same way that classical computers are used today to run the machine learning algorithms that power desktop and smartphone apps. “Your mom is not going to have a quantum computer,” Killoran says - but once a quantum solution shown to be scalable, it will become increasingly likely that one is processing her searches, clicks and taps.