Amazon Braket quantum computing service glues together three unrelated systems
Amazon is taking a “choice is good” approach to quantum, as different qubit designs are suited to different classes of problems.
Amazon unveiled Braket, a quantum computing service that represents a largely perfunctory entry in the quantum computing market.—Itit brings choice, certainly, as Amazon provides cloud access to D-Wave, IonQ, and Rigetti quantum computers—but practically extracting any value from any of these remains the task of the user.
To start, D-Wave, IonQ, and Rigetti represent three different strategies toward quantum computing: D-Wave offers a quantum annealer, IonQ uses trapped-ion qubits, and Rigetti is advancing a superconducting qubit co-processor. D-Wave’s quantum annealer is a frequent target of criticism in the industry for being less robust than other approaches—their offering relies on all problems being expressed as a quadratic unconstrained binary optimization (QUBO) problem. Relatively, this is skim milk compared to the full-fat designs of other computers.
SEE: Quantum computing: An insider’s guide (free PDF) (TechRepublic)
While D-Wave contends that QUBO has business value—and there are businesses using it—few circumstances appear in which a D-Wave annealer performs well in head-to-head comparisons with anything else, for the practical reason that programmers essentially need to write specifically for D-Wave.
As an aside, D-Wave also garners criticism for characterizing their quantum annealer as containing up to 2,000 qubits. Amazon’s announcement blog post, published Monday, indicates that “as I write this, the largest quantum computers contain about 50 qubits.”
IonQ has the distinction of having partnerships with AWS, and with Azure Quantum, announced at Ignite 2019 in Orlando last month. This likely bodes well for IonQ in terms of availability to developers at a minimum, considering the size and market position of AWS and Azure. Likewise, Rigetti’s partnership is likely to help in mindshare. While there is no guarantee that the hosted platform model will win out for quantum—particularly keeping in mind differences between hardware developers—that delivery model has worked out for practically everything else up to this point.
Alongside the unveiling of Bracket, the company is rolling out a Quantum Solutions Lab, that the company touts as allowing organizations “to tap into our own expertise and that of our consulting partners. Our goal is to work with you to find those practical uses, and to help you to build up your own ‘bench’ of qualified quantum developers.”
Similarly, the AWS Center for Quantum Computing, starting at Caltech, aims to “bring together researchers and engineers from Amazon with leading academic institutions in quantum computing, to develop more powerful quantum computing hardware and identify novel quantum applications with the goal of boosting innovation in science and industry.”
Update (December 4, 2019):
A previous version of this article indicated that D-Wave systems possess up to 5,000 qubits. D-Wave’s currently commercialized system – the 2000Q – has 2,000 qubits. The referenced 5,000 qubit system is slated for availability in 2020.
Further, a D-Wave spokesperson told TechRepublic that “a universal quantum system is on our roadmap.” On programming, D-Wave contends that “for any system, you need to formulate the problem in a way that the QPU understands. D-Wave, among others, has tools to make this easier by handling high- and low-level abstractions.”
After the publication of this article, the Amazon blog post language was altered to state “the largest gated-based quantum computers contain about 50 qubits.”
Image: DR JON HERAS, © DR JON HERAS/Science Photo Library/Corbis