A quantum computer the size of a PC graphics card? One Australian-German startup says they’re working a technology that may deliver just that within five years.
Today, of course, quantum computers are typically as large as mainframes. Yet the startup Quantum Brilliance has unveiled market-ready, diamond-based quantum computer only the size of a server rack module. They say their envisioned graphics card-sized devices by 2026 could find a home onboard satellites and autonomous vehicles.
Whereas classical computers switch transistors either on or off to symbolize data as ones or zeroes, quantum computers use quantum bits—qubits—that, because of the surreal nature of quantum physics, can exist in a state of superposition where they are both 1 and 0 at the same time. This essentially lets each qubit perform two calculations simultaneously.
If two qubits are quantum-mechanically linked, or entangled, they can help perform 2^2 or four calculations simultaneously; three qubits, 2^3 or eight calculations; and so on. In theory, a quantum computer with 300 qubits could perform more calculations in an instant than there are atoms in the visible universe.
A drawback of many quantum computers is that they demand temperatures colder than those found in deep space as well as complex systems to control them. As such, they today typically come in the form of cabinets the size of large, bulky mainframes—and dedicated to solving only the most difficult and otherwise intractable problems, perhaps accessed online via the cloud.
Now Quantum Brilliance says they have developed a commercially available quantum computer based on synthetic diamond that can operate at room temperatures. "It's the size of a desktop computer or a 19-inch rack," says Marcus Doherty, a quantum physicist and chief scientific officer at Quantum Brilliance.
Quantum Brilliance's technology is based on defects in diamonds that each result in a missing carbon atom in the crystals, with a rogue nitrogen atom located nearby. These so-called "nitrogen-vacancy centers" serve as qubits, and the diamond helps protect them from thermal disruptions and magnetic impurities, enabling operation at room temperatures.
"Diamond quantum computing has been around since 2001," Doherty says. "But it hit a barrier in 2015 in scaling beyond a handful of qubits. Quantum Brilliance is now overcoming this barrier."
Instead of quantum mainframes, Quantum Brilliance aims to build what it calls “quantum accelerators,” compact, robust quantum computers akin to graphics accelerators on personal computers. In March, Quantum Brilliance announced it would install its first quantum accelerators at the Pawsey Supercomputing Center in Perth, Australia, in June.
"Instead of a single large quantum computer with many qubits, one can think about small but many quantum computers that might not have as many qubits as a large quantum mainframe, but can still provide a quantum advantage for select tasks," Doherty says. The Pawsey Supercomputing Center's Quantum Pioneer Program is developing software for these quantum computers, he adds.
The first generation of Quantum Brilliance's quantum accelerators only hosts 5 qubits. However, “in five years time, it will be the size of a graphics card with 50-plus qubits,” Doherty says.
With 53 qubits, Google claimed it achieved a quantum advantage over classical computers, with its Sycamore quantum computer performing a computation in just 200 seconds that the company estimated would take the Summit supercomputer 10,000 years. “We predict that with 14 qubits, we will be able to out-perform a modern CPU processor in a desktop computer at the same task,” Doherty says.
One set of applications Quantum Brilliance imagines for its devices involves what it dubs massively parallel quantum computing, with many quantum accelerators working together on a problem.
For instance, with the kinds of molecular dynamics simulations often used to discover new drugs, “a quantum computer with a relatively small number of qubits can outperform an individual server in the rack of a supercomputer—it's a dramatic speedup,” Doherty says. “By adding many of these quantum computers together, you can then simulate a complicated chemical system, with each quantum computer simulating one molecule within a complex chemical system containing many molecules.”
Another set of applications for Quantum Brilliance's technology involves mobile quantum accelerators in what the company dubs edge quantum computing.
“Let's imagine you have a satellite with which you want to perform image or signal processing,” Doherty says. “Satellites gather a huge amount of signals or images, and there's often not enough computational power onboard for pre-filtering or processing of that data. Streaming huge amounts of data is also a problem, due to limited communications links. So one thing quantum computers do well is sort through combinatorial possibilities, which is exactly what is done in the feature identification and feature tracking in image and signal processing. Quantum computing can filter the workflow onboard a satellite or other constrained environments, such as autonomous vehicles.”
When it comes to autonomous vehicles and other autonomous systems, quantum accelerators “can also support decision making,” Doherty says. “Quantum computers are very good at explorations of possible sequences of events and identifying what are the most likely and what are the most dangerous sequences of events. So if you have a neural net or other simulator that can analyze a current situation and simulate likely possible futures, a quantum accelerator can search through that and figure out the most likely and most dangerous futures.”
Another application for edge quantum computing would involve natural language processing, “such as speech-to-text,” Doherty says. “Currently edge devices that perform speech to text have a lot of errors because the chips miss correlations between two bits of speech, two bits of a word. A quantum accelerator can help reduce those errors, so within the same timeframe, you could get a step change in accuracy.”
“One more application for the edge is with medical imaging, where there are constraints other than computational power limitations — for instance, there are constraints around privacy when it comes to transmitting medical information over the cloud,” Doherty says. “You can install a quantum computer on site for image processing with MRI and CT scans.”
Quantum mainframes may outperform quantum accelerators when it comes to running the kind of quantum algorithms designed to run only on quantum computers. When you have many small quantum computers working on such pure quantum algorithms, their computational power adds up linearly, but when you have one large quantum computer with the same number of qubits all entangled together, “you see an exponential growth in computational power,” Doherty says.
However, many applications for quantum computing actually involve quantum computers working together with classical computers on so-called hybrid quantum algorithms. In these cases, the fact that Quantum Brilliance's quantum accelerators may integrate more deeply with classical computers could reduce the total amount of time exchanging information between the two different kinds of computers “to outperform a single large quantum mainframe,” Doherty says.
Charles Q. Choi is a science reporter who contributes regularly to IEEE Spectrum. He has written for Scientific American, The New York Times, Wired, and Science, among others.