Quantum computing company D-Wave is setting ambitious targets for itself in the next few years. In a new roadmap called Clarity, the company has committed to becoming the first provider of both quantum annealers and gate model quantum computers – and if it delivers, it could make big waves in the industry.
Clarity effectively expands on D-Wave's current offering, which is limited to cloud-based superconducting quantum annealers, to start including a superconducting gate model quantum computer – similar to the ones developed by tech giants such as IBM and Google – in the next couple of years.
By 2024, according to D-Wave's roadmap, the first gate model preview quantum processing units (QPUs) will come online in the company's quantum cloud service, along with a suite of adapted tools and resources. The service will also include cross-platform solvers for users to tap both annealing and gate model processors within the same platform.
Quantum annealing and gate model quantum computers are two different approaches to quantum computing that researchers are currently putting to the test. It is still unclear whether one method is better than the other; rather, they both come with their own unique pros and cons.
Quantum annealing relies on processors that are much easier to control and operate, which means that they can already support thousands of qubits to tackle realistic problems faster than classical computers. But the class of problems that can be addressed by quantum annealers is limited by the design of the processor. Annealers are only suited to a specific set of optimization problems, which limits their capabilities.
On the other hand, gate model quantum computers could one day resolve a vast array of problems, ranging from simulating complex molecules to improving machine-learning models. The design of gate model quantum processors, however, is much more complex and still limited. Leading companies in the space, like IBM, can barely boast 100 qubits, meaning that those processors are unable to tackle problems of any significant size just yet – although this is expected to change in the next few years.
For the past 15 years, D-Wave has been making a name for itself by building superconducting quantum annealers. The method has enabled the company to pack an impressive number of qubits on its quantum processors: D-Wave's latest technology, called Advantage, counts 5,000 qubits.
This enables the company's customers, therefore, to use quantum computers to solve business problems, while gate model quantum computers can only be used experimentally for now. Canadian grocery chain Save-On-Foods, for example, is using D-Wave's services in a real-life setting, to manage in-store logistics more efficiently.
"We are quite good at solving optimization problems," Alan Baratz, the CEO of D-Wave, tells ZDNet. "We've done this in employee scheduling, manufacturing, shipping logistics, and more. But as we work with customers, we also see where the other opportunities are for quantum that we can't quite address today."
According to Baratz, the optimization use cases that annealers are best suited for account for 30% of the overall quantum application market. That's a "significant" portion, he argues, but is far from addressing the full spectrum of applications for the technology.
This is what prompted the company to kickstart a program to develop a gate model quantum computer that could tackle a wider range of problems – and specifically, a superconducting quantum computer.
Although different approaches exist, such as trapped ions or photonic systems, Baratz explains that superconducting systems were the most logical pick for the company.
"Our investment over the last years in superconducting annealing technologies can very much be applied to a scaled error-corrected gate model system," says Baratz. "It's not as different as you might think."
D-Wave will be using a type of annealer-inspired qubit that is different to the transmon qubits employed by other superconducting quantum computers. The qubits will be controlled similarly to those on the annealing processor, which Baratz says will eventually enable D-Wave to achieve high qubit density on a single chip.
Without committing to any specific dates, the Clarity roadmap shares the progress towards a general-purpose gate model QPU in five "phases". It starts with the development of a single qubit, using multi-layer qubit design and on-chip control devices – best practices that have also been developed in annealing quantum systems.
A target 60-qubit system will then be used to demonstrate a logical qubit, before increasing the count to a 1,000-qubit system to be fabricated on a single die, which will be configurable as up to four error-corrected logical qubits. To compare, IBM has pledged to build a 1,121-qubit superconducting quantum processor by 2023.
"We are still a few years away from our first gate-based system," says Baratz. "But our goal is to get a small error-corrected system out into the hands of users."
D-Wave's customers will then be able to use two different quantum methods to resolve specific business problems. For example, a pharmaceutical company could use the gate model quantum processor to assist with drug discovery, which is a simulation problem; and then switch to the quantum annealer to ensure patient trial optimization.
In other words, the Canadian company is setting itself to provide cross-platform services that could support a much wider variety of customer needs.
As gate model quantum computers become better and larger, however, it is easy to speculate that D-Wave's latest move might reduce the company's focus on quantum annealers; but Baratz is keen to stress that both approaches will be working in conjunction.
"Annealing quantum computing really is the approach you need to use to solve optimization problems. It is the quantum workhorse for that," says Baratz. "For that reason, we continue to be very focused on improving annealing systems into the future, increasing qubits and increasing connectivity."
D-Wave's Clarity roadmap, in effect, also comes with a dedicated "annealers" chapter. The company has announced, in line with its promise to regularly update its annealing processors, that it has improved the performance of Advantage to enable users to solve larger problems with greater precision, with better results reported 70% of the time on certain classes of problems.
A new solver, called the constrained quadratic model (CQM) solver, has also been released to simplify the process of expressing a problem in the quantum realm for users.
D-Wave's next annealing architecture, the Advantage 2 quantum system, is due to be released by 2024 with more than 7,000 qubits and 20-way connectivity. This will further expand the size of the problems that can be addressed by customers.
And for 2025 and beyond, D-Wave is still planning to focus on qubit connectivity and coherence in its annealers. In parallel, the company will be releasing tools and documentation, such as code examples or gate model simulators, to prepare developers to take their first steps in cross-platform quantum computing.