Taking Quantum Computing to the Edge

Much of the current buzz in quantum computing involves “quantum as a service” (QaaS). In this approach to the market, quantum mainframes with hefty resource or infrastructure requirements—such as the superconducting-circuit computers being developed by Google and IBM, or the trapped-ion computers of companies like IonQ—are set up as cloud-based installations packaged with accompanying software and services. Users then send their data to these machines over the internet to run quantum routines. A report earlier this year from Yole Intelligence predicted that QaaS would account for most of the rapid growth envisioned for quantum computing in the remainder of this decade.

At least one quantum computer developer, however, is betting on substantial business opportunities with a very different model. The deep-tech startup Quantum Brilliance, based in Australia and Germany, is looking to build quantum-computing hardware and software that can live closer to where customer data is actually being created. The company even hopes someday to integrate chip-scale “quantum accelerators” directly into edge devices for robotics, medical imaging, space-based applications and more—untethered from cloud-based quantum mainframes or data centers.

One key to this vision lies in the company’s choice of quantum bits (qubits): nuclear spins in diamond nitrogen–vacancy (NV) centers. These qubits—which are optically addressable and sport long coherence times—can operate at room temperature, an important attribute for potential edge-computing applications. And the company thinks it has a way of fabricating such qubits that’s better than conventional approaches, and ripe for scaling up.

Last month, Quantum Brilliance released a software development kit (SDK), dubbed Qristal, to enable potential customers to test out quantum algorithms tailored to its hardware for “real-world applications.” To learn more about where the company wants to go, OPN recently talked with its chief revenue officer, Mark Mattingley-Scott.

Beyond the cloud

One reason Quantum Brilliance is looking at markets beyond QaaS, according to Mattingley-Scott, has to do with data security. The “real business value” of quantum computing, he says, will come when enterprises conclude that it can have “a transformative character on some core value-creation process” in their industry. Examples might be risk arbitrage in the banking business, optimizing logistics in manufacturing, or selecting drug-development candidates.

One reason Quantum Brilliance is looking at markets beyond QaaS, according to Mattingley-Scott, has to do with data security.

Those kinds of applications, Mattingley-Scott points out, tend to revolve around sensitive governance, compliance or intellectual-property topics, personal-data security or other tricky areas. “Those are exactly the things that no company on Earth would ever want to put in the cloud,” he says. “So I believe the first commercially viable use cases with quantum computing will be exactly not in the cloud.”

Apples and oranges

Instead of a cloud-based architecture built around quantum mainframes, Quantum Brilliance is looking at smaller quantum devices, on a scale comparable to existing CPUs or GPUs. These could be integrated into existing supercomputing centers as “quantum accelerator” modules. They could also be deployed in distributed computing and in mobile and edge devices to perform quantum speedups of computing at the source.

Mattingley-Scott notes that such a model changes how one sizes up so-called quantum advantage—how, as he puts it, you compare “classical apples with quantum oranges.”

In the mainframe world, quantum advantage is usually defined as the point at which a quantum computer can take on tasks impossible for even the most powerful classical supercomputer. But in a system of distributed quantum accelerators, according to Mattingley-Scott, the relevant question for a business case becomes not whether quantum technology can beat the best supercomputer out there, but whether it can outperform a CPU or GPU of the same size, weight and power requirement as the quantum accelerator itself.

“That’s a much more tractable question,” he says. “It’s more about: I have [a certain number of] qubits; I know they’re in a size, weight and power that outperforms the classical alternative. Can I actually leverage that power?”

Making diamond NV centers at scale

For Quantum Brilliance, diamond NV centers are the platform for answering that question. NV centers are point defects in a diamond crystal lattice, where two carbon atoms are replaced by one nitrogen atom and a lattice vacancy. The nuclear spins clustered around the NV centers act as qubits that can be initialized and read out optically, via interactions with the electron spin of the NV center.

One reason for looking at NV centers for edge-device quantum computing has to do with their robustness to environmental perturbations. The nuclear-spin qubit “doesn’t care about a lot of the negative [environmental] influences” that can lead to noise in superconducting-circuit or trapped-ion computers, Mattingley-Scott says, and that require those installations to be cooled to cryogenic temperatures or to employ complex laser setups. Instead, diamond NV center qubits can operate at room temperature.

Beyond the intrinsic favorable properties of diamond NV centers, Quantum Brilliance also thinks it has a better way of making them. The conventional approach to creating these centers, Mattingley-Scott points out, involves “firing a nitrogen shotgun” at a piece of electronic-grade diamond, a nondeterministic process that also requires annealing afterward to “smooth out the mess.” In contrast, he says, Quantum Brilliance has developed a proprietary process, analogous to lithography in the semiconductor world, that allows the company to place the defects precisely at a predetermined location—and at angstrom scales.

Beyond the intrinsic favorable properties of diamond NV centers, Quantum Brilliance also thinks it has a better way of making them.

QDK and SDK

At present, Quantum Brilliance is funded by a combination of venture capital and government grants. Earlier this year, the company closed on a US$18 million seed round involving an investor group led by Breakthrough Victoria. And it’s also a partner in a US$17.5 million, three-year project funded by the German Federal Ministry of Education and Research, aimed at developing a demonstration of a scalable quantum computer based on diamond NV centers.

As to products, right now Quantum Brilliance offers what it calls a Quantum Development Kit (QDK). The QDK is a two-qubit, rack-mountable prototype that Mattingley-Scott says is designed to enable benchmarking of NV-center qubit behavior and to explore ways to integrate the company’s quantum accelerators with existing supercomputer centers. Around a year ago, one of the Quantum Brilliance units was installed in Australia’s Pawsey Supercomputing Research Center as an initial test.

More recently, last month’s launch of the company’s open-source Qristal SDK provides a platform for testing quantum algorithms that might run on Quantum Brilliance’s accelerators. This is especially relevant, says Mattingley-Scott, for those working in specialized areas, such as computational chemistry, that are poised to benefit from quantum speedups. “That’s where the uptake will happen—when it becomes useful for science or engineering or enterprise IT,” he maintains.

Pushing to the edge

Ultimately, Quantum Brilliance wants to take its accelerators to edge-computing devices in areas ranging from robotics to biomedicine to autonomous vehicles to aerospace—what Mattingley-Scott refers to as quantum computing at the source. Providing quantum speedups and energy savings to such devices is a huge opportunity, in his view. “I think there are 8 million officially recognized data centers on the planet,” he points out, but “there are 23 billion edge and IOT [internet of things] endpoints.”

Providing quantum speedups and energy savings to edge devices is a huge opportunity, in Mattingley-Scott’s view. “I think there are 8 million officially recognized data centers on the planet,” he points out, but “there are 23 billion edge and IOT [internet of things] endpoints.”

Getting the scale to address that huge market will require, among other things, a lot of miniaturization. One key, Mattingley-Scott thinks, will be Quantum Brilliance’s proprietary “atom-scale fabrication” technique for building diamond NV centers. He believes it will allow the company to ramp up the number of qubits on a chip-scale diamond substrate from current single-digit numbers to the hundreds—in much the same way, he says, that the semiconductor industry has, through incremental improvement in the past half-century, gone from a few thousand transistors in a single integrated circuit to many billions.

Parallel to fine-tuning this fabrication process, Quantum Brilliance will also need to miniaturize the control electronics, optics and other supporting technology to allow mass manufacturing. Those are all significant engineering challenges, Mattingley-Scott admits—but “nothing that hasn’t been done before.”

“So imagine we’re producing a billion of these a year; they cost a few cents each; we’ve got 100 or 200 qubits,” he says. “That is transformative technology. And that’s where we’re headed.”