At its Qubits 2021 conference today, D-Wave announced its product roadmap including several new items on both the hardware and the software fronts. These include:
- A performance update to their Advantage quantum annealing system which they now call Advantage Performance. This update still has the same 5000 qubits and qubit connectivity of 15 as the previously system, but it is using new materials and a modified fabrication process which provides it with higher yield, more usable qubits and higher performance than the previous system. This provides the system the ability to solve many larger, more complex problems with greater precision than its predecessor. The Advantage Performance system is available now and is already accessible via Amazon’s Braket service.
- The company is also developing a next generation quantum annealer that they are calling Advantage 2. This processor will have about 7000 qubits and it will have increased connectivity of 20 from the previous 15. This machine will use a new fabrication stack as well as a new topology. In addition, improvements will be made to reduce the noise and improve the coherence times of the qubits. This new machine is expected to be available in the 2023-2024 timeframe.
- The company also announced that it has started development of a gate based machine. It is important to understand that D-Wave is doing this to expand their product line rather than to replace their annealing technology. The company believes that the annealing approach will always be the best solution for optimization problems, but admits there are many non-optimization problems where a gate based machine would be a better approach. They also believe that many of the core technologies they have developed for a superconducting annealer can also be applied to a superconducting gate based machine. These include multilevel chip fabrication, on-chip controls, tunable qubits, etc. This machine will use a flux based superconducting qubit instead of the transmon qubits used by IBM, Google, and Rigetti and it will include error correction. Although the company did not specify the specific timeframe on when this machine would be available, they did describe a five phase development process they will be using to create it. They are initially working on a small demonstration of a single qubit and will follow that by creating a device of about 60 physical qubits that will be used with error correction algorithms to create one logical qubit. Several phases will occur after these steps, but the company indicated that this will be a long development effort taking several years.
- On the software side, the company introduced a new hybrid solver called a Constrained Quadratic Model. The company has been working to make their software more powerful and easier to use. The original hybrid solver they created could only handle binary variables which needed to be either 0 or 1. Last year, they introduced a Discrete Hybrid Solver that could handle variable that were part of a set. For example, a problem might restrict the problem to a set of four colors such as red, blue, yellow, and green. This latest solver can now handle variables which can be any integer value. It also makes it easier to specify constraints to a users problem so they don’t have to manually program the constraints in themselves.
- Longer term, D-Wave will continue to evolve their software. They discussed future upgrades to their solver software that would allow it to handle continuously variable parameters. Also, they are planning on providing their Leap software with cross-platform capabilities that will allow a user to use the same interface for solving a problem on either an annealer, gate-based processor or a classical processor. They are planning to provide vertical software solutions that would be applicable to specific use cases and be very easy to use. In addition they also discussed providing a simulator for their future gate based machine.
D-Wave was originally founded in 1999 and has a wealth of issued patents as well as core technology related to superconducting implementations. They have many clients developing proof-of-concept (POC) demonstrations using their machines and indicated that a few of them have reached production. Their solutions should not be ignored in the race to demonstrate quantum advantage or obtaining revenue from providing quantum cloud services. Additional information about their roadmap is available in a news release posted on their website here and a blog article posted on Medium located here.
October 5, 2021