On September 2nd Xanadu announced it would offer cloud access to its gate-based photonic quantum computers. Xanadu is a Canadian quantum technology company founded in 2016 and obtained $32 million Series A financing in mid-2019. Additional grants from DARPA and Sustainable Development Technology Canada (SDTC) brought its total investment to $45 million.
Xanadu is the first company to offer cloud access to photonic quantum computers. For now, customers will be able to use either an 8-qubit or a 12-qubit machine. Rafal Janik, responsible for Xanadu cloud development, told Moor Insights & Strategy that Xanadu plans to put a 24-qubit processor online soon.
Robert Niffenegger is a Ph.D. and a member of the Trapped Ion and Photonics group at MIT Lincoln Laboratory. I asked him to weigh in on the Xanadu announcement. “Xanadu allowing users to access their photonic circuits via the cloud, is very impressive indeed, a first! I think they have plenty of room to increase their chip complexity quickly. Shor’s algorithm has been demonstrated on photonic circuits on a small scale before, so having a scalable path towards larger photonic circuits may be enough for the niche applications that quantum computers may fill in the future.”
The entire ecosystem advances
I believe this announcement advances the overall progress of quantum computing. It is a step toward a time when quantum computers will perform usefully and make groundbreaking computations.
Xanadu’s announcement gives researchers and customers cloud access to four different quantum computing technologies. Superconducting, trapped ion, quantum annealing (not gate based) are offered by other companies and on Amazon Braket and Microsoft Azure cloud platforms. For now, photonic quantum computers are provided only on Xanadu’s cloud.
The most common gate-based quantum computing systems are superconducting and trapped ion. Companies using superconducting qubits are IBM, Rigetti, and Google. Honeywell and IonQ offer access to trapped-ion quantum computers.
“We believe that photonics offers the most viable approach towards universal fault-tolerant quantum computing with Xanadu’s ability to network a large number of quantum processors together. We are excited to provide this ecosystem, a world-first for both quantum and classical photonics,” said Christian Weedbrook, Xanadu Founder and CEO, according to Xanadu’s press release. “Our architecture is new, designed to scale-up like the Internet versus traditional mainframe-like approaches to quantum computing.”
Why Xanadu is different
Most people have some familiarity with commonly used qubits. These qubits can exist in three states, a quantum state of one or zero, or a superposition of both values. A measurement causes the superposition to collapse and reveals that the qubit is either a one or a zero.
Integrated Photonics – How it Works
Xanadu uses a special type of qubit based on squeezed light and a concept called continuous variables. Instead of only three states like other qubits, conceptually, if you continue to add more and more states of differing values, you will eventually have a continuous set of numbers for sampling. Then, once you make a measurement, instead of being limited to a state of one or a zero, you will get a number like 2.8.
Despite the complexity of Xanadu’s photonic chip, in simple terms, it is a programmable system of entangled photons. The chip turns classical photons from laser pulses into a non-classical state of photons called a squeezed state. After a programmable interferometer performs operations on the squeezed states, the photons leave the chip where external detectors perform the readout. All applications require classical post-processing on the readout.
Niffenegger said that without quantitative loss or fidelity numbers, it is difficult to estimate how far Xanadu will be able to push its technology. He’s hopeful that cloud researchers will start running benchmarks on the system to see what performance is possible.
Machine Learning, AI, and other tools
According to the press release, developers will use Xanadu’s open-source tools available on Github. These tools include Strawberry Fields, its cross-platform Python library to simulate and execute programs on quantum photonic hardware, and PennyLane, its software library for quantum machine learning, quantum computing, and quantum chemistry.
Xanadu has stated that PennyLane is the leading QML framework, and it has plugins for all publicly available hardware.
Access to Xanadu
A select group of enterprise customers and partners are currently using Xanadu’s 8-qubit and 12-qubit photonic quantum computers. The users include students and researchers from universities worldwide and the Department of Energy researchers from Oak Ridge National Labs and Oak Ridge Leadership Computing Facility.
Once Xanadu provides an API key, a user can submit programs using strawberry fields (a hardware tutorial is available). A job is then submitted, queued, and executed in its order of submission.
Unlike IBM that allows open access to all users of its cloud of 28 quantum machines, gradually increasing users is common practice with startups and prototype technologies.
According to Rafal Janik, Xanadu has seen an overwhelming response to its service. “We are focusing our onboarding efforts on institutions right now, but we are quickly moving to include individual contributors as well,” he said. “As for open access, it’s not about control or limit; it’s just about how quickly we can onboard people.”
Future of Xanadu and quantum computing
Building a universal gate-based, fault-tolerant quantum machine with millions of qubits is the ultimate goal of almost every quantum computing scientist. It is generally believed that goal is a decade or more away.
Surprisingly, Xanadu’s roadmap projects it will reach a million qubits within 5-10 years. As mentioned before, and in the analyst notes, the quality of qubits is always an important factor.
Photonic quantum computing has made a great deal of progress over the past five years. As demonstrated by Xanadu, operations on individual photon squeezed states can now be done with precision using programmable interferometry. Improvements in integrated platforms and detectors have also helped advance the technology.
Once a machine like that becomes available, it will be capable of solving problems that are too difficult or even impossible for a classical computer to solve. A useful quantum computer with a million qubits will be a game-changer for environmental, scientific, medical, and pharmacological applications.
Even scaling quantum computers to a thousand high quality, error-corrected qubits is currently a major scientific challenge. We know what is required; we don’t know yet how to do it.
Photon qubits have the advantage of maintaining quantum states for a very long time. Photons are inconceivably small fields that are emitted from distant stars and travel for thousands and even billions of years before reaching our eyes still intact as photons. A good example is the Lyman-alpha blob. A photon from the blob is still polarized in its original state when it reaches earth after traveling for 11.5 billion years.
Admittedly, Xanadu has a very ambitious technology roadmap, but it is based on a promising technology. According to Christian Weedbrook, founder and CEO, the company anticipates doubling the number of qubits every six months. If 24 qubits come online later this year, that means the company should deliver a cloud machine with close to 100 qubits by the end of 2021. However, raw qubit counts alone are an incomplete measurement of quantum computers. See my analyst notes.
New photonic generations
Xanadu’s current Gaussian boson sampling chip is a prototype used for both its own and customer research. Scaling is among its present limitations. Niffenegger confirmed this when he observed that the existing platform is more like an ASIC [Guassian Boson Sampling is a specialized chip] than a CPU. “While it may be able to scale up the number of paths possible for its superposition states,” he said, “the layout of the photonic circuit limits reconfigurable gates for the paths.”
For Xanadu to reach an intermediate benchmark of 1000 or more qubits, it will need a new architecture. Photonic research is robust, and many workable options are available. It is likely Xanadu already has its next-generation architecture under development, or perhaps even on the shelf and ready to deploy.
However, moving beyond a thousand qubits to a million qubits will require more research and additional generations of its quantum processor. That leap is a greater challenge with many unknowns. There will be many engineering and science problems to overcome before a million qubits can become a reality.
- Once a universal fault-tolerant quantum computer is built, it is possible that many technical advantages currently enjoyed by photonics will have been solved and engineered into other technologies as well.
- Unless a major new technology breakthrough occurs, quantum computing progress made over the next five years will likely reveal which qubit type is superior and most scalable.
- For the foreseeable future, quantum computing will rely on the cloud. From a customer’s perspective, until quantum computers are on its premises, it will make little difference if a computer operates at room temperature or if it requires a dilution refrigerator or if it needs a metal trap.
- There will be some emerging applications that require a compact, room temperature, low power consumption quantum computer. This paradigm will be especially attractive to the military.
- From a research standpoint, room temperature operation provides a large engineering benefit and allows for faster development and easier scaling. Both of which will be important to some customers.
- As mentioned in my previous articles, qubit counts alone are not accurate reflections of a quantum computer’s computational power. According to Rafal Janik, there is no direct way to compare the continuous variable approach to IBM’s or Honeywell’s quantum volume. He volunteered to review it with the Xanadu team.
- The addition of Xanadu’s open access to its photonic processor on the cloud contributes to the entire quantum computing ecosystem. Xanadu’s processors provide researchers and developers with novel approaches that are unique to solve problems in finance, quantum chemistry, machine learning, and graph analytics. Xanadu deserves a big thanks for that contribution.
- It is logical to expect that Xanadu will eventually appear on one or both of the Amazon Braket or Microsoft Azure Quantum platforms.
- Companies such as IBM and Honeywell using superconducting and trapped-ion technologies also have aggressive research programs in place and roadmaps that lead to universal, fault-tolerant gate-based quantum computers.
Note: Moor Insights & Strategy writers and editors may have contributed to this article.
Disclosure: Moor Insights & Strategy, like all research and analyst firms, provides or has provided paid research, analysis, advising, or consulting to many high-tech companies in the industry, including 8×8, Advanced Micro Devices, Amazon, Applied Micro, ARM, Aruba Networks, AT&T, AWS, A-10 Strategies, Bitfusion, Blaize, Calix, Cisco Systems, Clear Software, Cloudera, Clumio, Cognitive Systems, CompuCom, Dell, Dell EMC, Dell Technologies, Diablo Technologies, Digital Optics, Dreamchain, Echelon, Ericsson, Extreme Networks, Flex, Foxconn, Frame, Fujitsu, Gen Z Consortium, Glue Networks, GlobalFoundries, Google (Nest-Revolve), Google Cloud, HP Inc., Hewlett Packard Enterprise, Honeywell, Huawei Technologies, IBM, Ion VR, Inseego, Intel, Interdigital, Jabil Circuit, Konica Minolta, Lattice Semiconductor, Lenovo, Linux Foundation, MapBox, Mavenir, Marseille Inc, Mayfair Equity, Meraki (Cisco), Mesophere, Microsoft, Mojo Networks, National Instruments, NetApp, Nightwatch, NOKIA (Alcatel-Lucent), Nortek, Novumind, NVIDIA, ON Semiconductor, ONUG, OpenStack Foundation, Oracle, Poly, Panasas, Peraso, Pexip, Pixelworks, Plume Design, Portworx, Pure Storage, Qualcomm, Rackspace, Rambus, Rayvolt E-Bikes, Red Hat, Residio, Samsung Electronics, SAP, SAS, Scale Computing, Schneider Electric, Silver Peak, SONY, Springpath, Spirent, Splunk, Sprint, Stratus Technologies, Symantec, Synaptics, Syniverse, Synopsys, Tanium, TE Connectivity, TensTorrent, Tobii Technology, Twitter, Unity Technologies, UiPath, Verizon Communications, Vidyo, VMware, Wave Computing, Wellsmith, Xilinx, Zebra, Zededa, and Zoho which may be cited in this article.