In 1994, Peter Shor, an American mathematician working at Bell Labs, published a paper with a wonky title and earth-shaking implications. In “Polynomial-Time Algorithms for Prime Factorization and Discrete Logarithms on a Quantum Computer,” Shor showed how, using a hypothetical quantum computer, you could efficiently solve certain types of mathematical problems that overwhelmed even the most powerful classical computer systems—eliminating many, many steps, and cutting the calculation time exponentially.
His theory was mind-blowing, though at the time only physicists, mathematicians, and a certain breed of computer scientist noticed. One of the most widely used digital encryption protocols, RSA, relies on finding the prime factors of very large numbers—a task so hard for classical computers that cracking such encryption could take several years. This is because as the numbers to be factored get longer, the computing power required grows exponentially. But Shor demonstrated that by leveraging the weirdness of quantum mechanics, you could arrive at a solution in minutes or even seconds. Existing encryption would be rendered obsolete.
The quantum computer that Shor envisioned didn’t exist yet. But his paper helped launch a far-flung effort to build one. But building such a computer would not just transform digital encryption—it would have profound implications for physics, chemistry, materials science, drug development, and other areas where complex problems involving huge numbers of variables confound conventional computers.
It would be seven years until a team of researchers from IBM used an early type of quantum computer to demonstrate that Shor’s algorithm worked—if only to solve an easy version of the problem: factoring the number 15. Subsequent teams have executed Shor’s algorithm on different kinds of quantum computers, for numbers up to . . . 21! Factoring numbers of practical significance—not to mention performing the other miracles that quantum evangelists promise—requires bigger, better quantum computers than now exist. But after decades as a perennial technology of the future, quantum computing has hit an inflection point.
The number of quantum computing deals soared more than 700% from 2015 through 2023, according to PitchBook data, and total deal value grew tenfold to $1 billion. Governments around the world have made quantum computing a strategic national defense priority. As of February 2024, the U.S government had invested $3 billion in quantum computing projects, plus an additional $1.2 billion from the National Quantum Computing Initiative. China, meanwhile, has reportedly invested some $15 billion in quantum computing efforts. Big Techs including IBM, Intel, Google, Microsoft, AWS, and Baidu all have substantial quantum computing programs, envisioning applications in fundamental physics, chemistry, materials science, drug development, finance, climate modeling, and the training and optimizing of AI models.
But in the past two months, a stealthy Palo Alto-based startup called PsiQuantum has leapfrogged these big names to become a surprise leader in the space. With a total funding of $700 million from the likes of BlackRock, Founders Fund, Playground Global, and Microsoft’s venture arm, the eight-year-old startup was tapped in April to build a utility-scale computer in Brisbane, Australia. Today, it was named as the anchor commercial tenant in a massive new quantum research campus on Chicago’s South Side. With over a half billion in funding from the state of Illinois, the city, and DARPA, the Department of Defense’s advanced research wing, the ambitious project aims to establish the United States as a leader in quantum technologies, ahead of worrisome adversaries—namely China.
PsiQuantum’s approach is radically different from that of its competitors. It’s relying on cutting-edge “silicon photonics” to manipulate single particles of light for computation. And instead of taking an incremental approach to building a supercomputer, it’s focused entirely on coming out of the gate with a full-blown, “fault tolerant” system that will be far larger than any quantum computer built to date. The company has vowed to have its first system operational by late 2027, years earlier than other projections. A big question now is: Will that be soon enough?
Qubits to Computers
Perhaps the only thing harder than building a quantum computer is explaining how it works. The physicist Richard Feynman, who shared the 1965 Nobel Prize in physics for his fundamental work in quantum electrodynamics, famously said, “If you think you understand quantum mechanics, then you don’t understand quantum mechanics.”
Being hard to grasp “is intrinsic to quantum physics,” says Peter Shadboldt, PsiQuantum cofounder and chief science officer. “The mathematics of quantum physics fits on a postcard. But there are thousands of pages of popular science trying to make that mathematics into a rewarding analogy. That property of quantum mechanics to resist analogy carries through to the computer.”
But let’s try imagining a coin being flipped. In classical computing, information is encoded in binary fashion, as 1s or 0, in magnetic bits. In this example, the coin has landed and shows either “heads” or “tails.” Quantum computers, on the other hand, store information in qubits—quantum bits—which can be both 1 and 0 at the same time. This state, called superposition, is like a coin that’s in the air but hasn’t yet landed—occupying varying degrees of “headness” or “tailness” until it lands.
The ability of qubits to exist in multiple states simultaneously, and to stitch themselves together through quantum entanglement, could enable sufficiently sophisticated quantum computers to run multidimensional algorithms with mind-melting speed—basically crunching all possible variables and combinations at once, rather than sequentially. “A really simple way to put it is that quantum physics introduces new rules to what you can do with information, so you have new strategies that you can deploy,” says Shadbolt.
There are numerous ways to make qubits. IBM, Google, and others use conductive metals to create so-called superconducting circuits that function as artificial atoms. Companies such as Honeywell and IonQ use “trapped” ions of natural atoms as qubits. Berkeley, California-based Atom Computing uses neutral atoms. PsiQuantum and Toronto-based Xanadu both build photons based on light, using single photons and “squeezed light” photons, respectively.
Making a qubit is just the start, though. Building a functional quantum computer involves much more. First, qubits must remain stable, or “cohere,” in a state of superposition for long enough to make a computation. You need ways to get them to interact with each other, and ways to control and measure them. You also need to mitigate and correct for the “noise” inherent to quantum systems. “The transistor on your laptop goes wrong way less than one in a trillion times,” says Shadbolt. “Everyone’s qubits today go wrong one in a thousand times, maybe one in 10,000 times, if you’re really lucky.”
In order to create a “fault tolerant” quantum computer that can produce useful results despite subatomic chaos, most companies plan to devote a substantial number of qubits—thousands and thousands of them—solely to correcting errors. “Our bet was no utility short of the order of a million qubits,” says PsiQuantum CEO and cofounder Jeremy O’Brien. And that means they have to be scalable through mass manufacturing.
Seeing the Light
As an undergraduate studying quantum mechanics in Perth, Australia in the mid-1990s. O’Brien read an article on quantum computing and Shor’s algorithm that changed everything for him. “It was far and away the wildest thing that I’d ever encountered,” he says. “It made the case that if we could build a quantum computer, harnessing these weird and wacky things that we were learning, you’d bring about a revolution comparable to fire, agriculture, or steam power.” But after exploring superconductors and other ways of making quantum computers, he says, “I had a full-scale crisis. I didn’t see how we were ever going to get to a million qubits using any of these approaches.”
Pursuing a Ph.D. at the University of New South Wales in Sydney, he fell under the spell of photonics, a well-developed technology widely used in fiber-optic networking systems in the telecom industry. In a post-doc position at the University of Queensland, he studied with one of the inventors of that approach, and in 2009, published a research paper that described how encoded photons of light could be used to perform quantum computations. In 2015, he and three like-minded colleagues at the University of Bristol, in England—Terry Rudolph, Mark Thompson, and Pete Shadbolt—founded PsiQuantum to pursue the all-or-nothing goal of creating a utility-scale, fault-tolerant photonics-based quantum computer.
For many years, the quantum computing industry had been driven by an approach called NISQ (noisy intermediate-scale quantum), says O’Brien, “building incrementally bigger systems and then trying to find commercial applications for them.” Google had demonstrated “quantum supremacy” with a 53-qubit computer in 2009, performing a specific math problem much faster than a classical supercomputer. “That’s a genuinely important scientific accomplishment,” says O’Brien. “But it’s basically generating noise beyond what any conventional system could produce.” At the scale you need to solve meaningful quantum problems, that noise would be deafening.
O’Brien and his founders had to discard the old playbook. “If you think that the goal is the top of the Empire State Building and you’ve got a 10 meter ladder, then you build a hundred meter ladder to get you closer to the top,” says O’Brien. “But when you realize that actually the goal is the moon, it doesn’t really matter whether you’ve got a 10-meter ladder or a 100-meter ladder. You need a rocket. So, what sort of fuel could you use for a rocketing? How would you build a prototype engine and test that fuel? What’s the fuselage? We grew and built a research center focused on solving the whole thing end to end.” Nearly 20 years later, the rocket is taking shape.
In January, PsiQuantum’s R&D engineering manager, Matthew House, gave a tour of the company’s headquarters and lab, housed in a nondescript Palo Alto low-rise across the street from the headquarters of EV maker Rivian. There, in a warren of cluttered workbenches divided by clear-plastic, clean-room strip curtains, company scientists kicked the tires on disparate parts of the system, surrounded by computer monitors, racks of testing equipment, and small black cryostats used to cool electronic components.
Qubits made with superconducting circuits or ions are fragile—prone to interference from the tiniest waves of energy in the universe. To minimize interference from “thermal noise,” they must be cooled to millikelvin temperatures, tiny fractions of a degree above absolute zero. Qubits made with photons can operate at room temperature. “Photons have the unique property that they do not experience interference from any other signals,” says House. The only parts of PsiQuantum’s system that require cooling are components used to detect photons, but those can operate at relatively balmy temperatures a few whole degrees above absolute zero, which requires much less energy. That’s another big advantage for scaling. “The cooling power that will be needed to scale up an IBM machine, for instance, will require facilities that just don’t exist today,” House says.
Another edge PsiQuantum has over many competitors is its ability to leverage tools and processes from the semiconductor industry. “We try to use as few novel processes as possible,” says O’Brien. From the beginning, O’Brien was “obsessed” with building photonic qubits on standard 300 millimeter silicon wafers—using standard optical components like waveguides, phase shifters, and polarizers to create integrated circuits that could produce and channel individual light particles, rather than electrons. Since 2017, Psi Quantum has partnered with tier-1 semiconductor manufacturer Global Foundries to develop an innovative “silicon photonic” chip, produced at GF’s facility in Malta, New York, as well as a proprietary electronic control chip, made at the GF foundry in Dresden, Germany.
“Everyone likes to argue that their qubit’s better than everyone else’s,” says O’Brien. “But actually the question is what stands between you and a hundred, or a million qubits. That’s about manufacturability—can you make millions to billions of components and test them along the way? Almost all of the time and money and energy that we’ve deployed has been in solving the scaling challenges.”
PsiQuantum has tested tens of thousands of chips, with the aim of reducing optical loss—the problem of photons going AWOL. This is about refining not only chip architecture, but also the manufacturing process. “How do you make the material that much more perfect when you etch waveguides?” House explains. “What’s the right shape of the wave guides? The roughness of the walls of the waveguide on the chip are a big thing. We’re getting down to the nanometer and working through the foundry to perfect those processes to get lower and lower optical loss.”
PsiQuantum isn’t just developing chips, though. Those chips need to be integrated into a big system to do something useful. “The thing about systems is everything needs to work, even the boring bits,” says O’Brien. One major challenge is “packaging”—figuring out how the necessary wires and fibers attach to devices and connect different parts of the system. A quantum computer will require a bunch of control electronics that need to interact with the qubits but may not have compatible working environments. What’s more, PsiQuantum and companies like IBM look to scale further by connecting parallel quantum processors through quantum networks that do not yet exist.
And for all the advantages of the company’s approach, it has unique hurdles. “Each modality has its pros and cons,” says Jussi Sainiemi, a partner at Voima Ventures, a deep tech venture firm in Helsinki. “Photonics-based qubits are likely more scalable, but the current technological challenges are likely on qubit control and entanglement.” In other words: light is pretty hard to manipulate precisely. Jay Gambetta, VP at IBM Quantum, raises similar concerns: “I can engineer superconducting circuits to have strong coupling. Photons generally don’t want to interact.”
PsiQuantum has developed tricks to overcome these challenges, but the ultimate test will be putting all the pieces together. Unlike IBM, which introduced its 1,121-qubit Condor quantum processor in December 2023, PsiQuantum has yet to assemble a complete system. Nonetheless, PsiQuantum’s reported progress and dozens of patents have given investors—and now governments—enough confidence that they’re offering the startup the funding and space to build.
Collapsing the Superposition
Moonshots typically draw on support from government. Quantum computing is no exception. In 2023, the U.S. government boosted its R&D spending on quantum technologies to $932 million, more than double the $449 million it spent in 2019. The Defense Quantum Acceleration Act of 2024, now in committee, calls for another $800 million “to accelerate the adoption and implementation of quantum information science technology” within the DOD.
Driving these investments is concern that a foreign adversary like China could leverage advances in quantum computing to gain a strategic and commercial advantage that would be hard to make up. China is reported to have put more than $15 billion of public funds into quantum computing, and scientists there last year reported a speed record, using the JiuZhang 3 photonic quantum computer to solve a complex Gaussian boson sampling (GBS) in milliseconds, versus an estimated 20 billion years for the most powerful known supercomputer. Luckily, this isn’t useful for code-breaking. But it demonstrates very advanced capabilities.
In January 2024, PsiQuantum was selected for the second phase of a Defense Advanced Research Projects Agency (DARPA) project that seeks approaches to quantum computing that could be capable of achieving utility-scale operation much faster than conventional predictions. Microsoft was the only other company to make the cut. In April, the Australian government tapped PsiQuantum to build a full-size version of its quantum computer, on the order of a million qubits, in Brisbane, in a deal worth about $620 million USD. The company expects the facility, which will be the size of a small data center and use proprietary cooling cabinets, to be operational by the end of 2027. (IBM’s quantum roadmap, by comparison, aims for a fault-tolerant computer in 2029.) Australia, the U.K., and the U.S. have committed to cooperating on the development of non-nuclear strategic technologies including AI and quantum computing through a trilateral security framework known as AUKUS.
But the biggest boost to PsiQuantmum’s work came today, with the announcement by Illinois Governor JB Pritzker that PsiQuantum will be the anchor tenant of an ambitious new quantum campus called the Illinois Quantum and Microelectronics Park, to be built at the site of a long-vacant former U.S. Steel plant site on Chicago’s South Side. There, the company will occupy a nearly 300,000-square foot facility that will house the largest, most powerful quantum computer in the country, if not the world. The campus will be the physical center for the Quantum Proving Ground program, a state-federal partnership with DARPA, which will invest up to $140 million. The state will spend about $300 million on a cryogenic cooling facility and development of the site, plus additional matching funds. And the city, county, and state will offer another $500 million in tax incentives over time.
Chicago mayor Brandon Johnson says PsiQuantum has been “a tremendous partner with my administration every step of the way. Jeremy O’Brien and I are like the Michael-Scotty combination of the ’90s.” The public-private partnership positions the city as a tech hub, and shows, Johnson says, that “we are open for business.” It also enables “families and working people to be able to live in these communities that were left behind when the mills left.”
The new quantum campus is projected to create thousands of jobs and have a total economic impact of as much as $60 billion. “When you look at the steel industry and all that it meant for global growth economically, that happened right here in Chicago,” says Johnson. “Quantum computing has the same fuel to do what the steel industry did, and I believe it could be even more transformational.”
In quantum mechanics, superpositions collapse into a single, “classical” reality once they are observed. From many possible realities, only one gets to be “real.” It is too soon to say that PsiQuantum is the one way forward for useful quantum computing. But the whole industry will certainly now be watching the company’s next moves. “We’ve got a very good idea and then we’ve been very serious about executing and refining our blueprint for eight-plus years,” says O’Brien. “I think basically we see the escape velocity moment here—then, we can go to the moon and beyond.”