In 2009, Jeremy O’Brien, a professor at the University of Bristol, published a research paper describing how to repurpose on-chip optical components originally developed by the telecom industry to manipulate single particles of light and perform quantum operations.
By 2016, based on the earlier photonic research, O’Brien and three of his academic colleagues, Terry Rudolph, Mark Thompson, and Pete Shadbolt, created PsiQuantum.
The founders all believed that the traditional method of building a quantum computer of a useful size would take too long. At the company’s inception, the PsiQuantum team established its goal to build a million qubit, fault-tolerant photonic quantum computer. They also believed the only way to create such a machine was to manufacture it in a semiconductor foundry.
PsiQuantum first popped up on my quantum radar about two years ago when it received $150 million in Series C funding which upped total investments in the company to $215 million.
That level of funding meant there was serious interest in the potential of whatever quantum device PsiQuantum was building. At that time, PsiQuantum was operating in a stealth mode, so there was little information available about its research.
Finally, after receiving another $450 million in Series D funding last year, PsiQuantum disclosed additional information about its technology. As recently as few weeks ago, a small $25 million US government grant was awarded jointly to PsiQuantum and its fabrication partner, GlobalFoundries, for tooling and further development of its photonic quantum computer. Having GlobalFoundries as a partner was a definite quality signal. GF is a high-quality, premiere fab and only one of the three tier one foundries worldwide.
With a current valuation of $3.15 Billion, PsiQuantum is following a quantum roadmap mainly paved with stepping stones of its own design with unique technology, components, and processes needed to build a million-qubit general-purpose silicon photonic quantum computer.
Classical computers encode information using digital bits to represent a zero or a one. Quantum computers use quantum bits (qubits), which can also represent a one or a zero, or be in a quantum superposition of some number between zero and one at the same time. There are a variety of qubit technologies. IBM, Google, and Rigetti use qubits made with small loops of wire that become superconductors when subjected to very cold temperatures. Quantinuum and IonQ use qubits formed by removing an outer valence electron from an atom of Ytterbium to create an ion. Atom Computing makes neutral atom spin qubits using an isotope of Strontium.
Light is used for various operations in superconductors and atomic quantum computers. PsiQuantum also uses light and turns infinitesimally small photons of light into qubits. Of the two types of photonic qubits – squeezed light and single photons – PsiQuantum’s technology of choice is single-photon qubits.
Using photons as qubits is a complex process. It is complicated to determine the quantum state of a single photon among trillions of photons with a range of varied frequencies and energies.
Dr. Pete Shadbolt is the Co-Founder and Chief Strategy Officer of PsiQuantum. His responsibilities include overseeing the application and implementation of technology and scientific-related policies and procedures that are vital to the success of PsiQuantum. After earning his PhD in experimental photonic quantum computing from the University of Bristol in 2014, he was a postdoc at Imperial College researching the theory of photonic quantum computing. While at Bristol, he demonstrated the first-ever Variational Quantum Eigensolver and the first-ever public API to a quantum processor. He has been awarded the 2014 EPSRC “Rising Star” by the British Research Council; the EPSRC Recognizing Inspirational Scientists and Engineers Award; and the European Physics Society Thesis Prize.
Dr. Shadbolt explained that detecting a single photon from a light beam is analogous to collecting a single specified drop of water from the Amazon river’s volume at its widest point.
“That process is occurring on a chip the size of a quarter,” Dr. Shadbolt said. “Extraordinary engineering and physics are happening inside PsiQuantum chips. We are constantly improving the chip’s fidelity and single photon source performance.”
Just any photon isn’t good enough. There are stringent requirements for photons used as qubits. Consistency and fidelity are critical to the performance of photonic quantum computers. Therefore, each photon source must have high purity, proper brightness, and generate consistently identical photons.
The right partner
GlobalFoundries facility in Essex, Vermont
When PsiQuantum announced its Series D funding a year ago, the company revealed it had formed a previously undisclosed partnership with GlobalFoundries. Out of public view, the partnership had been able to build a first-of-its-kind manufacturing process for photonic quantum chips. This manufacturing process produces 300-millimeter wafers containing thousands of single photon sources, and a corresponding number of single photon detectors. The wafer also contains interferometers, splitters, and phase shifters. In order to control the photonic chip, advanced electronic CMOS control chips with around 750 million transistors were also built at the GlobalFoundries facility in Dresden, Germany.
Every quantum qubit technology has its own set of advantages and disadvantages. PsiQuantum chose to use photons to build its quantum computer for several reasons:
Another major advantage of photon qubits worth highlighting is the ability to maintain quantum states for a relatively long time. As an example of light’s coherence, despite traveling for billions of years, light emitted by distant stars and galaxies reaches earth with its original polarization intact. The longer a qubit can maintain its polarized quantum state, the more quantum operations it can perform, which makes the quantum computer more powerful.
Why start with a million qubits?
“We believed we had cracked the code for building a million-qubit quantum computer,” Dr. Shadbolt said. “Even though that’s a huge number, the secret seemed simple. All we had to do was use the same process as the one being used to put billions of transistors into cell phones. We felt a large quantum computer wouldn’t exist in our lifetime unless we figured out how to build it in a semiconductor foundry. That idea has been turned into reality. We are now building quantum chips next to laptops and cell phone chips on the GlobalFoundries 300-millimeter platform.”
According to Dr. Shadbolt, PsiQuantum’s custom fabrication line has made much progress. Surprisingly, building a million-qubit quantum machine in a foundry has many of the same non-quantum issues as assembling a classical supercomputer, including chip yields, reliability, high-throughput testing, packaging, and cooling – albeit to cryogenic temperatures.
“From the time that our first GlobalFoundries announcement was made until now, we’ve produced huge amounts of silicon,” Dr. Shadbolt said. “We’ve done seven tapeouts in total and we’re now seeing hundreds and hundreds of wafers of silicon coming through our door. We are investing heavily in packaging, assembly systems, integration, and fiber attachment to ensure the highest efficiency of light flowing in and out of the chip.”
PsiQuantum is performing a great deal of ongoing research as well as continually improving the performance of photonic components and processes. In addition to high-performance optical components, the technologies that enable the process are also very important. A few enablers include optical switches, fiber-to-chip interconnects, and bonding methods.
“We have greatly improved the efficiency of our photon detectors over the last few tapeouts at GlobalFoundries,” Dr. Shadbolt explained. “We’re constantly working to prevent fewer and fewer photons from being lost from the system. We also have driven waveguide losses to extremely low levels in our recent chips.
“There is much innovation involved. Our light source for single photons is a good example. We shine laser light directly into the chip to run the single photon sources. The laser is about a trillion times more intense than the single photons we need to detect, so we must attenuate light on that chip by a factor of about a trillion.”
Dr. Shadbolt attributes PsiQuantum’s manufacturing success to GlobalFoundries. From experience, he knows there is a significant difference between a second-tier foundry and a first-tier foundry like GlobalFoundries. Building chips needed by PsiQuantum can only be built with an extremely mature manufacturing process.
“PsiQuantum has two demanding requirements. We need a huge number of components, and we need those components to consistently meet extremely demanding performance requirements. There are very few partners in the world who can reliably achieve something like this, and we always knew that partnering with a mature manufacturer like GlobalFoundries would be key to our strategy.”
The partnership has also been beneficial for GlobalFoundries because it has gained additional experience with new technologies by adding PsiQuantum’s photonic processes to the foundry.
The end is in sight
According to Dr. Shadbolt, the original question of whether large numbers of quantum devices could be built in a foundry is no longer an issue as routinely demonstrated by its output of silicon. However, inserting new devices into the manufacturing flow has always been difficult. It is slow and it is very expensive. Nanowire single photon detectors are an example of a development that came directly from the university lab and was inserted into the manufacturing process.
PsiQuantum’s semiconductor roadmap only has a few remaining items to complete. Since a million qubits won’t fit on a single chip, the quantum computer will require multiple quantum processor chips to be interconnected with optical fibers and facilitated by ultra-high-performance optical switches to allow teleportation and entanglement of single photon operations between chips.
“What remains is the optical switch,” Dr. Shadbolt said. “You might ask why photonic quantum computing people have never built anything at scale? Or why they haven’t demonstrated very large entangled states? The reason is that a special optical switch is needed, but none exists. It must have very high performance, better than any existing state-of-the-art optical switch such as those used for telecom networking. It’s a classical device, and its only function will be to route light between waveguides, but it must be done with extremely low loss and at very high speed. It must be a really, really good optical switch.”
If it can’t be bought, then it must be built
Implementing an optical switch with the right specs is a success-or-fail item for PsiQuantum. Since a commercial optical switch doesn’t exist that fits the application needs, PsiQuantum was left with no choice but to build one. For the past few years, its management has been heavily investing in developing a very high-performance optical switch.
Dr. Shadbolt explained: “I believe this is one of the most exciting things PsiQuantum is doing. Building an extremely high-performance optical switch is the next biggest thing on our roadmap. We believe it is the key to unlocking the huge promise of optical quantum computing.”
PsiQuantum was founded on the belief that photonics was the right technology for building a fault tolerant quantum machine with a million qubits and that the proper approach was based on semiconductor manufacturing. In contrast to NISQ quantum computers, the founders wanted to avoid building incrementally larger and larger machines over time.
Considering the overall process needed to build a million-qubit quantum computer, its high degree of complexity, and the lack of proven tools and processes to do it with, PsiQuantum has made amazing progress since it first formed the company.
It established a true partnership with one of the best foundries in the world and produced seven tapeouts and funded a half dozen new tools to build a first-of-its-kind wafer manufacturing process, incorporating superconducting single photon detectors into a regular silicon-photonic chip.
And today, it is answering yet another challenge by building an optical switch to fill a void where the needed product doesn’t exist.
It is no surprise that an ultra- high-performance optical switch is a key part of PsiQuantum’s plans to build a scalable million qubit quantum computer. Other quantum companies are also planning to integrate similar optical switching technology to scale modular QPU architectures within the decade. The high-performance optical switch PsiQuantum is developing could someday connect tens of thousands of quantum processing units in a future multi-million qubit quantum data center. As a standalone product, it could also be a source of additional revenue should PsiQuantum choose to market it.
Once the optical switch has been built, it will then need to be enabled into GlobalFoundries’ manufacturing flow. That is the last step needed to complete PsiQuantum’s foundry assembly process and then it will be ready to produce photonic quantum computer chips.
But even with a complete end-to-end manufacturing process, significantly more time will be needed to construct a full-blown fault-tolerant quantum computer. It will remain for PsiQuantum to build complete quantum computers around chips produced by GlobalFoundries. For that, it will need a trained workforce and a location and infrastructure where large qubit photonic quantum computers can be assembled, integrated, tested, and distributed.
Based on the amount of post-foundry work, development of the optical switch, and assembly that remains, and assuming no major technology problems or delays occur, I believe it will be after mid-decade before a photonic quantum computer of any scale can be offered by PsiQuantum.
I’ll wrap this up with comments made by Dr. Shadbolt during our discussion about the optical switch. I believe it demonstrates why PsiQuantum has been, and will continue to be successful:
“Even though the optical switch will obviously be a very powerful generic technology of interest to others, we are not interested in its generic usefulness. We are only interested in the fact that it will allow us to build a quantum computer that outperforms every supercomputer on the planet. That is our singular goal.”
Paul Smith-Goodson is Vice President and Principal Analyst for quantum computing, artificial intelligence and space at Moor Insights and Strategy. You can follow him on Twitter for more current information on quantum, AI, and space.
Note: Moor Insights & Strategy writers and editors may have contributed to this article.
Moor Insights & Strategy, like all research and tech industry analyst firms, provides or has provided paid services to technology companies. These services include research, analysis, advising, consulting, benchmarking, acquisition matchmaking, and speaking sponsorships. The company has had or currently has paid business relationships with 8×8, Accenture, A10 Networks, Advanced Micro Devices, Amazon, Amazon Web Services, Ambient Scientific, Anuta Networks, Applied Brain Research, Applied Micro, Apstra, Arm, Aruba Networks (now HPE), Atom Computing, AT&T, Aura, Automation Anywhere, AWS, A-10 Strategies, Bitfusion, Blaize, Box, Broadcom, C3.AI, Calix, Campfire, Cisco Systems, Clear Software, Cloudera, Clumio, Cognitive Systems, CompuCom, Cradlepoint, CyberArk, Dell, Dell EMC, Dell Technologies, Diablo Technologies, Dialogue Group, Digital Optics, Dreamium Labs, D-Wave, Echelon, Ericsson, Extreme Networks, Five9, Flex, Foundries.io, Foxconn, Frame (now VMware), Fujitsu, Gen Z Consortium, Glue Networks, GlobalFoundries, Revolve (now Google), Google Cloud, Graphcore, Groq, Hiregenics, Hotwire Global, HP Inc., Hewlett Packard Enterprise, Honeywell, Huawei Technologies, IBM, Infinidat, Infosys, Inseego, IonQ, IonVR, Inseego, Infosys, Infiot, Intel, Interdigital, Jabil Circuit, Keysight, Konica Minolta, Lattice Semiconductor, Lenovo, Linux Foundation, Lightbits Labs, LogicMonitor, Luminar, MapBox, Marvell Technology, Mavenir, Marseille Inc, Mayfair Equity, Meraki (Cisco), Merck KGaA, Mesophere, Micron Technology, Microsoft, MiTEL, Mojo Networks, MongoDB, MulteFire Alliance, National Instruments, Neat, NetApp, Nightwatch, NOKIA (Alcatel-Lucent), Nortek, Novumind, NVIDIA, Nutanix, Nuvia (now Qualcomm), onsemi, ONUG, OpenStack Foundation, Oracle, Palo Alto Networks, Panasas, Peraso, Pexip, Pixelworks, Plume Design, PlusAI, Poly (formerly Plantronics), Portworx, Pure Storage, Qualcomm, Quantinuum, Rackspace, Rambus, Rayvolt E-Bikes, Red Hat, Renesas, Residio, Samsung Electronics, Samsung Semi, SAP, SAS, Scale Computing, Schneider Electric, SiFive, Silver Peak (now Aruba-HPE), SkyWorks, SONY Optical Storage, Splunk, Springpath (now Cisco), Spirent, Splunk, Sprint (now T-Mobile), Stratus Technologies, Symantec, Synaptics, Syniverse, Synopsys, Tanium, Telesign,TE Connectivity, TensTorrent, Tobii Technology, Teradata,T-Mobile, Treasure Data, Twitter, Unity Technologies, UiPath, Verizon Communications, VAST Data, Ventana Micro Systems, Vidyo, VMware, Wave Computing, Wellsmith, Xilinx, Zayo, Zebra, Zededa, Zendesk, Zoho, Zoom, and Zscaler. Moor Insights & Strategy founder, CEO, and Chief Analyst Patrick Moorhead is an investor in dMY Technology Group Inc. VI, Dreamium Labs, Groq, Luminar Technologies, MemryX, and Movandi.
Moor Insights & Strategy founder, CEO, and Chief Analyst Patrick Moorhead is an investor in dMY Technology Group Inc. VI, Dreamium Labs, Groq, Luminar Technologies, MemryX, and Movand