D-Wave’s 500-Qubit Machine Hits the Cloud

When quantum-computing pioneer D-Wave releases its next-generation Advantage2 system in 2023 or 2024, the company expects its 7,000-qubit machine to be the most powerful quantum computer of its kind in the world. Now D-Wave is making an experimental prototype of Advantage2 immediately available for use over the cloud.

Classical computers switch transistors either on or off to symbolize data as ones or zeroes. In contrast, quantum computers use quantum bits, or “qubits.” Because of the strange nature of quantum physics, qubits can exist in a state called superposition, in which they are essentially both 1 and 0 at the same time. This phenomenon lets each qubit perform two calculations at once. The more qubits are quantum mechanically linked, or entangled, within a quantum computer, the greater its computational power can grow, in an exponential fashion.

The standard approach toward building quantum computers, called the gate model, involves arranging qubits in circuits and making them interact with each other in a fixed sequence. In contrast, D-Wave—based in Burnaby, B.C, Canada—has long focused on what are called annealing quantum computers. Quantum cousins of classical annealing computers, these machines find a lowest energy state by slowly cooling it down—in much the same way that metals and crystals are sometimes tempered so as to minimize imperfections. Quantum annealing machines, then, start off with a set of qubits whose interactions at their lowest energy state, called the ground state, represent the correct answer for a specific problem the researchers programmed it to solve.

“Given the early positive results, we wanted to get it into the hands of developers and researchers now for exploration and learning.”
—Emile Hoskinson, D-Wave

The ideal application for annealing quantum computers may be solving optimization problems, says Emile Hoskinson, an experimental physicist and the director of quantum-annealing products at D-Wave. These seek to find the best answer from all possible solutions, such as mapping the fastest route from point A to point B.

For instance, imagine trying to find the lowest point on a vast landscape covered in hills and valleys. A classical computer might start at a random spot on the surface and look around for a lower spot to explore until it cannot walk downhill anymore. This approach can often get stuck in a “local minimum,” a valley that is not actually the very lowest point on the surface.

On the other hand, annealing quantum computers could make it possible to start at many spots on the surface at the same time, reducing the chance of becoming trapped in a local minimum. They can even essentially tunnel through a hill to see if there is a lower valley beyond it, or share data from multiple spots to find patterns that might lead to deeper points.

Founded in 1999, D-Wave bills itself as the world’s first commercial supplier of quantum computers. The company has long proved controversial, with many critics over the years questioning whether its machines are any more powerful than regular computers.

Nevertheless, D-Wave has claimed its share of high-profile clients over the years. It sold its first quantum computing system, the 128-qubit D-Wave One, to Lockheed Martin in 2011, and shipped its 512-qubit D-Wave Two to NASA’s Quantum Artificial Intelligence Lab—launched in partnership with Google and the Universities Space Research Association—in 2013.

When D-Wave’s Advantage2 comes online next year or the year after, it will be the world’s most powerful annealing quantum computer, Hoskinson says. Now the company is releasing an experimental prototype of Advantage2, with all the core functionality of the full-scale product available for testing.

“Our broad portfolio of enterprise customers—such as Volkswagen, Save-on-Foods, Denso, Toyota, BBVA, NEC, Accenture, and Lockheed Martin—have built hundreds of early quantum applications in diverse areas such as resource scheduling, mobility, logistics, drug discovery, portfolio optimization, manufacturing processes, and much more,” Hoskinson says. “We expect the Advantage2 can be used to address problems with even greater complexity. When we talk about addressing a broader swath of problems, we mean the improved technology can tackle problems that are larger, more complex, and that cover a wider range of use cases across different verticals. It will address the same problems as before, but better and faster.”

The prototype holds more than 500 superconducting flux qubits. The new design of the qubits enables what D-Wave calls its Zephyr topology, which supports 20-way interqubit connectivity, up from 15 in the company’s prior generation, Advantage.

“An analogy to illustrate the importance of connectivity is a social network, in which influence, and the complexity of interactions, grows with the number of connections between nodes,” Hoskinson says. “Complexity can mean problem size, such as the number of variables, constraints, and so on. It can also mean commercial and business application context—for example, if a particular business needs a fast time-to-solution because of business demand, such as changing schedules, supply-chain dependencies, and so on, then that makes the problem more complex. We see Advantage2 solving problems both better and faster, addressing both of these types of complex use cases.”

The new qubit design also supports a higher energy scale, which makes the qubits less vulnerable to disruption from thermal fluctuations. This in turn reduces error rates in quantum computation.

In tests, when the prototype’s qubits were arranged in their standard “native” topology, it found better solutions in up to 89 percent of cases when compared with Advantage, the company notes. When it came to problems requiring greater connectivity between qubits, multiple qubits can get linked together in a strategy called embedding, and Advantage2’s greater interqubit connectivity led the prototype to find better-quality solutions than Advantage in up to 82 percent of such problems.

“Many commercially interesting problems require embedding,” Hoskinson says. “With the new Advantage2 Zephyr topology and increased energy scale, we see performance improvements for both native and embedded problems.”

D-Wave did not originally plan to make its prototype available publicly, “but given the early positive results, we wanted to get it into the hands of developers and researchers now for exploration and learning,” Hoskinson says. “It’s about learning from our community to maximize commercial application performance as we build towards the final full-scale Advantage2 quantum computer.”

Hoskinson notes that D-Wave is also working on a new fabrication process that should lead to dramatically less noise in the company’s quantum computers, increasing their chances of finding high-quality solutions.

“While the prototype was developed in our current rapid-development fabrication stack, the eventual Advantage2 product will be produced in an all-new lower-noise stack,” he says. “We have early results for the new stack that show a seven-times reduction in low-frequency flux noise, a three-times reduction in integrated flux noise, and an order of magnitude reduction in high-frequency flux noise. This will go a long way toward further improving performance in the full Advantage2 system.”

D-Wave made the prototype available over its Leap quantum cloud service in June.