Quantum computers have been traditionally classified into either quantum annealing processors or gate level quantum processors. But up until now, there has only been one organization, D-Wave, pursuing quantum annealers with the rest of the industry (Microsoft, IBM, Rigetti, Ionq, etc.) pursuing the gate-level versions. However, that may be changing with a recent launch by the U.S. Government agency IARPA (Intelligence Advanced Research Projects Activity) of a five year program called QEO (Quantum Enhanced Optimization) to develop quantum annealing technology far beyond the current levels of today’s technology.
The QEO program will team up academic groups from USC, MIT, Harvard, CalTech, UC Berkeley, University College London, University of Waterloo, Saarland University, and Tokyo Institute of Technology together with commercial companies Lockheed Martin and Northrop Grumman, and complemented by the Government test and evaluation teams at NASA Ames Research Center and Texas A&M, and the Government Furnished Capability teams at MIT Lincoln Labs and MIT; all to design a quantum annealer with a verified ten-thousand time speed increase over the best state-of-the-art classical solution.
A guiding philosophy behind this program is to discover the optimal path to enhancement: to develop novel and far more powerful architectures with a smaller numbers of very high quality qubits, rather than a much larger but less powerful architecture employing lower quality qubits. The Program goals only call for 100 qubits, but according to Dr. Karl Roenigk, QEO Program Manager at IARPA, the proposed 100 qubit QEO machines are anticipated to significantly exceed the computational power commercial capability today at well beyond 1000 qubits. However, to be fair to others working on quantum annealers, they will not be standing still and plan to make significant architectural improvements in their future machines that will also improve the quality of their qubits. (See the previous article we published on D-Wave’s roadmap here.)
The goals of this program will be to build on high-coherence qubits and develop significant improvements in qubit connectivity, coupling coefficient dynamic range, higher-rank Ising couplings, error mitigation, non-stoquastic terms, and other features. These improvements will make it easier to map problems onto the machine and provide more precise and accurate solutions to meet the 10,000X speed-up goal. A key aspect of this program will be to leverage the 3D flip-chip flux qubits developed in earlier programs at MIT Lincoln Labs and MIT, as well as all the leading advances in classical methods and quantum annealing that additional QEO team members at NASA, ETH Zurich, and Texas A&M have achieved in the past few years. For details on the technical goals of the program, please refer to Tables 1 and 3 of the BAA (Broad Agency Announcement) which can be found here.
Although some folks may believe that quantum annealers may start fading out as gate level machines become available, this may not be the case. Due to the initial availability of the D-Wave machines, folks working on quantum annealing have a significant head start towards developing libraries and applying quantum annealing to real world problems and this may prove to be a significant advantage. Quantum annealers will likely continue to be a better choice for certain types of problems that require optimization or sampling even after gate-level machines start becoming available commercially. And the competition will help up everyone’s game. Like it always does.
So the question that remains is how or when this technology will be commercialized. Assuming the program meets its technical goals there should high interest from potential users to make this available commercially. Northrup Grumman is the lead for the design, fabrication, system integration, and experimental testing portion of this program, so they might be in the best position to publicly offer an advanced quantum annealer based on this program. Or another possibility might be a classical computing company that wants to expand their product line and license this technology or partner with some of the organizations in this program to bring this technology to market. There would still be a lot of issues if they decide to go down this path, but certainly one can see a lot of possibilities. It will be interesting to see how this develops.