This New Quantum Processor Is Made of Light

Experts are still unclear on what material will make the best quantum computer, but one obscure candidate is gaining traction: infrared light. Earlier this month, quantum computing startup Xanadu, based in Toronto, Canada, put two of its so-called photonic quantum computers on the cloud—the first commercially available devices of their kind.

Xanadu’s computers each consist of a silicon chip about the size of a thumbnail, with 8 and 12 infrared laser beams shining onto them, respectively. To execute an algorithm, the computer carefully puppeteers the beams to reflect, combine, and interact in a controlled way. You can think of the chip as an abacus of sorts, where the chip solves math problems by manipulating laser beams instead of wooden beads. To do arithmetic on an abacus, you move the beads around according to a set of rules and count the resulting number of beads in each row to get the answer. Similarly, in Xanadu’s quantum computer, you interact with each beam of light and find the answer to your problem by counting the resulting number of photons in each beam. But unlike beads, these photons follow the rules of quantum mechanics, enabling much more complex math than addition and subtraction. Quantum computing experts think that these devices should be able to solve certain business-relevant math problems faster than conventional computers, although they have yet to conclusively demonstrate these claims.

Currently, the material that a quantum computer is made of predetermines what that machine is good at. Just as it’s easier to build boxy structures with Legos and amorphous blobs with Play-Doh, a quantum computer made of light can solve certain math problems more readily compared to a quantum computer made of superconducting circuits (like Google’s), and vice versa. “The set of problems that you can solve on our cloud is literally different from anyone else,” said Xanadu CEO Christian Weedbrook, who is a physicist by training.

Researchers working at Xanadu have identified various use cases in business and chemistry where quantum computers might offer more capability than conventional computing. For example, their devices should be able to solve currently intractable instances of the so-called densest subgraph problem, where the user is trying to identify which node in a network has the most connections. “You can think of it as finding the biggest influencer in a social media network,” said Weedbrook. Selfies and hashtags aside, the densest subgraph problem also has applications in biology and medicine, where, for example, researchers are trying to understand how complicated networks of proteins interact in the human body. Weedbrook said that Xanadu aims to demonstrate its devices’ quantum advantage in a business application in the next year.

Anyone can apply for time on the computers, although Xanadu is prioritizing researchers at government labs, multinational corporations, and multi-user institutions. Over 250 people have already applied to use Xanadu’s devices, according to Weedbrook. Once approved, a user receives a digital token to connect to the computer via Xanadu’s software platform, Strawberry Fields, where they can run Python-based code on the quantum device remotely.

Xanadu’s hardware design offers other benefits. For example, it is relatively straightforward to connect Xanadu’s computers using conventional fiber optics because they use infrared light, the same as existing telecom infrastructure. This could simplify quantum internet schemes, proposed for at least a decade from now, for linking multiple quantum devices over the world. In addition, photonic quantum computers should theoretically be able to operate at room temperature, although currently Xanadu still has to cryogenically cool its photon counter.

However, because of hardware limitations, Xanadu’s machine is currently equipped to execute only a specific set of algorithms, said physicist Giulia Ferrini of Chalmers University of Technology in Sweden, who researches algorithms for photonic quantum computers and is not affiliated with Xanadu. The ultimate goal for the quantum computing industry is to build a so-called universal quantum computer, which can solve a broad class of problems. Xanadu’s computer “is a first step,” she said.

Because it uses light, Xanadu’s computer exploits quantum properties of photons distinctive from other commercially available quantum computers. For example, the computer exploits Heisenberg’s uncertainty principle, which says that nature is inherently imprecise, in a unique way. If you measure a photon’s speed precisely, then you sacrifice precision in knowing its location, for example. To perform its computations, a Xanadu device controls which properties of light are uncertain, a technique known as “squeezing.” In addition, quantum mechanical weirdness is also evident when the computer counts the photons in each laser beam. Before the computer counts the photons, the number of particles in the beam is undefined—the light actually contains a superposition of 1, 10, and 20 photons, for example. Like Schrödinger’s cat, which is only alive or dead when you look at it, the number of photons in the beam is only set when you count them.

Xanadu’s design contrasts with all other commercially available quantum computers, which are based on the so-called qubit, or quantum bit, architecture. As opposed to a classical bit, which represents information as 1 or 0, a quantum bit is a superposition of 1 and 0, like a coin flipping in the air is neither heads nor tails but some probability of both. But Xanadu’s computer doesn’t use 1’s and 0’s at all. Instead, it uses properties of light—the intensity and the time a beam is first emitted—that can take on any value.

To stay consistent with the lingo in the rest of the industry, Xanadu refers to each light beam on its chip as a qubit, but technically they are not qubits at all. In quantum computing jargon, each light beam is something called a “qumode,” short for “quantum mode,” where “mode” is a fancy word for light whose wavelike properties exhibit specified patterns. Such a design is known as a continuous-variable architecture.

Xanadu’s devices mark the debut of a new way to do math using quantum mechanics. “I think this quantum computer will spark further interest in the broader quantum technology community towards these continuous-variable architectures,” said Ferrini. Xanadu’s devices may lend legitimacy to other researchers who are trying to build continuous-variable quantum computers using microwave radiation instead of infrared light.

Xanadu is looking to its customers to figure out what role these computers can play in the future. Researchers at Oak Ridge National Laboratory, for example, have purchased access to Xanadu’s devices in order to design the next generation of supercomputers. They think that future supercomputers could contain a quantum computing chip—a “QPU,” or quantum processing unit—that is dedicated to performing specialized tasks fast, said Oak Ridge quantum computer scientist Travis Humble. In their efforts to design a QPU, Oak Ridge has bought access to other commercial quantum computers as well, such as devices made by IBM, D-Wave, Rigetti—“as many as we can get our hands on,” said Humble.

To be sure, Xanadu’s new devices are still very much part of quantum computing’s adolescence, an era that experts call NISQ, for Noisy Intermediate-Scale Quantum Computing. NISQ machines work in limited ways, and it’s still unclear how the machines will integrate into existing technology and what purpose they will serve. “It’s like when you hit a growth spurt when you’re young, and you outgrow all your clothes and get awkward,” said Humble. “Quantum computing is off to its next plateau in development.”