PsiQuantum, a little-known quantum computing startup, recently received $150 million in a Series C funding round that closed earlier this year. The funding was led by Atomico, with participation from other frontier tech VCs, like Founders Fund, growth investors like Blackrock and Ballie Gifford, and strategic partners like Microsoft’s M12 Ventures. Including the latest funding, PsiQuantum has raised a total of $215 million since its founding.
The company was founded in 2016 by British professor Jeremy O’Brien and three other academics, Terry Rudolph, Mark Thompson, and Pete Shadbolt. In just a few years, they have quietly grown the company from a few employees to a robust technical staff of more than 100.
Compared to today's modest quantum computing capabilities, PsiQuantum's elevator pitch for investors sounds like a line from a science fiction movie. O'Brien not only says he is going to build a fault-tolerant quantum computer with a staggering one million qubits, he also says he is going to do it within five years. O'Brien's technology of choice for this claim is silicon photonics, which uses particles of light called photons to perform quantum calculations. Theoretically, photons behave as both waves and particles, but that's a subject for another article. Quantum computing technologies in use today are primarily superconductors and trapped ion. However, there is plenty of research that shows photonics holds a lot of promise.
A look at qubit technologies
While classical computers use magnetic bits to depict ones and zeros for computation, quantum computers use a variety of other technologies to make quantum bits called qubits.
PsiQuantum's objective to build a quantum computer with a million qubits is a colossal undertaking. For perspective, today's biggest and best quantum computers have less than 100 qubits. Even that number stretches the limits of present-day quantum science. Google recently achieved quantum supremacy by performing a difficult computational task in a matter of seconds that would have taken a classical computer thousands of years to complete. Moreover, it only took a mere 54 qubits for that historic achievement.
I asked Robert Niffenegger, a research scientist at MIT Lincoln Labs, for his thoughts on PsiQuantum and its goal of a million qubits. Niffenegger said, "By setting a goal of a million qubits they emphasize that scale and integration are the only path forward and flaunt the fact that existing nano-photonics based on CMOS fabrication technologies is able to fabricate thousands of optical components on a single chip. However, even if they had very high-performance photonics on a single photonic chip the size of a wafer, that would at best get you maybe thousands of qubits."
Superconducting qubits, the most commonly used technology for quantum computing, are the foundation of those built by Google, Intel, IBM, and Rigetti. The devices are basically small coils fabricated on chips that resemble those found in classical computers.
Optical and SEM images of a transmon qubit
NATIONAL INSTITUTE OF STANDARDS AND TECHNOLOGY
Quantum effects kick in when the coils are cooled to a few degrees above absolute zero and become superconductors. At that temperature, current flows resistance free in a clockwise or counterclockwise direction and represents either a one or a zero or a superposition of everything between one and zero.
Superconducting qubits can be manufactured using existing chip fabrication techniques. A few drawbacks with superconducting qubits include:
1.) they lose their quantum states quickly, limiting the number of sequential calculations that can be performed on a problem.
2.) they can only connect to their nearest qubit neighbor. Several connections are needed to reach a distant qubit, much like steppingstones placed across a stream. Unfortunately, those extra connections slow down calculations and limit the complexity of problems that can be solved.
Trapped ion qubits
Trapped ions are the oldest qubit technology, dating back to the 1990s. Honeywell and IonQ are the most prominent commercial users of trapped ion qubits. Atomic clocks also use trapped-ion technology.
String of 14 trapped and entangled ions
UNIVERSITY OF INNSBRUCK
Honeywell and IonQ use qubits formed from an isotope of rare-earth metal called ytterbium, although other materials can also be used. Precision lasers remove an outer electron from an atom of ytterbium to create an ion. Lasers are also used like tweezers to move ions around. Once in position, oscillating voltage fields hold the ions in place.
Compared to superconducting qubits, ions maintain their quantum states for a very long time. The longer a quantum state can be maintained, the more complex of a computation can be performed. Honeywell leveraged this attribute and recently announced it would have the world's most powerful quantum computer when its 8-qubit trapped ion machine is released in a few months.
Using light to make qubits
Instead of using coils or ions as qubits, PsiQuantum plans to build its quantum computer using single particles of light, called photons.
A photon can be vertically polarized to represent a one, or horizontally polarized as a zero, or even diagonally polarized to represent a superposition of both one and zero.
PsiQuantum's secret sauce is a 2009 research paper written by its founder, Jeremy O’Brien. This research and other quantum tricks allow qubits to be encoded by photons traveling at the speed of light.
A significant advantage of photon qubits is that they maintain their quantum states for a very long time. Photons from distant stars and galaxies travel for thousands and even billions of years before reaching our eyes. A good example is a Lyman-alpha blob. Photons from the blob are still polarized in their original state when they reach earth after traveling for 11.5 billion years.
In addition to PsiQuantum, several other research groups are trying to figure out how to scale up photonic computers to more qubits. However, unlike PsiQuantum, none have a stated goal of a million qubits.
A photon qubit is very small. It has a wavelength of about one-millionth of a meter (μm). In some ways, a photon's small size is an advantage, but in other ways its size creates obstacles that PsiQuantum must overcome to reach a million qubits.
Photons travel at the speed of light (after all, they are particles of light ), and that makes them very hard to control.
Imagine trying to manipulate something the size of a virus as it zips past you at a speed of 300,000,000 meters per second. Unlike superconducting qubits that are fixed in place and ions that remain stationary, photons are always in motion. It will be challenging to juggle the state of millions of blazing fast photons while simultaneously trying to read, control, and manipulate each one of them.
Observations and conclusions
Here are my thoughts and conclusions from an analyst's perspective:
1. PsiQuantum claims it will be able to do things in five years that many quantum experts predict will take 7 to 10 years or more to accomplish.
2. Silicon photonics appears to be a promising technology to build a quantum computer capable of solving complex problems that are far beyond the capabilities of classical supercomputers. Niffenegger also shared his thoughts on this: "I believe they [PsiQuantum] do have a path to becoming the 'supreme' heavyweight champion of the quantum crown, and I think that if they publish some smaller-scale demonstrations, then other people will start to believe it too."
3. Having Microsoft as a strategic investor will provide PsiQuantum with access to many critical resources needed to build a silicon photonic quantum computer.
4. Error correction is a significant quantum computing problem for every present-day qubit technology. According to publicly available information, PsiQuantum places a great deal of emphasis on error correction. That means a large portion of the million-qubit goal will likely be devoted to monitoring and correcting errors. For today's error prone qubits, it is estimated that thousands of error correcting qubits are required for every computation qubit.
5. Even if PsiQuantum is only able to produce a thousand error corrected qubits, they will have created a fault-tolerant quantum computer that might change the world. It could create new drugs, design new materials, model DNA, and make thousands of other major scientific, medical, and commercial breakthroughs.
6. Remember, Microsoft had similar optimistic goals as PsiQuantum when it began research on a topological quantum computer. That was a decade ago. Tangible results today: zero.
PsiQuantum has made some outrageous claims that I believe will end in one of two ways. Either the company revolutionizes the space or flames out like few startups have flamed out flushing its investors time and money down the drain.
Note: Moor Insights & Strategy writers and editors may have contributed to this article.