The year 2019 was a busy year in the quantum community with a lot of new developments and announcements.  We are sure that 2020 will be just as busy, if not more so, and expect continued advances.  We have seen some of the roadmaps that folks in the industry have discussed so with our intrepid 20/20 vision (pardon the pun!) we will describe some of the developments that we expect to see this year.


Hardware providers will continue to make advances in both qubit count, qubit quality and new technologies in 2020.  Things we expect to see include:

  • In September 2018, Rigetti announced a new architecture they call Aspen starting with a 16 qubit chip, advancing to an intermediate density of 32 or 64 qubits with a large version of 128 qubits in their roadmap.  In December they announced their 32 qubit version, called Aspen-7, which will run on both Rigetti’s QCS as well as Amazon’s Braket cloud services.  In 2020, we expect that they will announce availability of the 128 qubit version on these services too.
  • In October 2019, Google announced that they successfully completed their quantum supremacy experiment with their 53 qubit Sycamore chip.  Since then they have hinted at various industry conferences that they are working on a 57+ qubit version of this chip with some improved qubit quality metrics.  We think the reason for the “+” is that the chip may have 59 or 60 qubits in the design, but a few might not work due to mechanical or yield failures.  However, they have indicated that a large focus of their 2020 activity will be to improve the gate fidelities.  In particular, their current Sycamore had an average 0.62% two-qubit simultaneous gate error and they would like to get that down to 0.1%.
  • IBM has publicly declared their intention to double what they call Quantum Volume every year.  Qubit Volume is a metric that can provide a general description of a machines power that takes into account the number of qubits, the quality of the qubits and other factors.  When they announced their 20 qubit IBM Q System One in 2019 they indicated that they had achieved a Quantum Volume factor of 16 with the design.  In September 2019, they announced a 53 qubit machine, but have not yet announced an associated Quantum Volume metric for it. We suspect they are still calibrating it and tuning it up for peak performance and they are waiting until this is done before announcing a quantum volume.  In any case, we do expect them to hit a goal of a doubling of Quantum Volume to 32 in 2020 either with this 53 qubit machine or perhaps something else they may have in development.
  • Another new superconducting entrant in 2020 will be Quantum Circuits Inc. (QCI).  They have announced a partnership with Microsoft that will serve as their cloud provider.  Not many details are public about their quantum computer, but we expect to hear more in the coming year.
  • 2020 should be a big year for D-Wave with the production release of their 5000+ qubit Advantage architecture based upon their Pegasus chip.  This architecture should bring substantial improvements in performance due to the increase number of qubits,improved coherence times, and better qubit connectivity.

    We are aware that D-Wave is working on a technology called nonstoquastic Hamiltonian which allows qubits to be coupled with two degrees of freedom rather than the current one degree of freedom.  This technology can substantially improve the problem solving capability of the quantum annealing machines.  Although we do not expect this to be included initial in the 2020 Advantage machines, we would expect them to announce a roadmap indicating that this technology will be incorporated in follow-on machines in the 2021-2022 time frame.
  • 2020 will see the first public availability of several cloud based ion trap machines from IonQ, Honeywell, and Alpine Quantum Technologies (AQT).  IonQ and Honeywell will be partnering with Microsoft and AWS to provide cloud access, but it is not yet known if they will also offer their own cloud services. AQT has some interesting possibilities because their ion trap hardware is now programmable via both Google’s Cirq as well as IBM’s Qiskit (see below).
  • We also expect to see cloud availability of photonic technologies from Xanadu.  They use a different type of element called qumodes rather than qubits.  Qumodes are continuously variable elements and may have some advantages for certain computations.  Xanadu current has a 12 qumode machine in the lab and expect to be offering one with 50+ qumodes by the end of 2020. Another distinguishing feature of photonic technologies is that they do not require the expensive dilution refrigerators that are needed by the other technologies.
  • Another new technology introduction that we expect to see in 2020 is the first cloud based computer based upon spin qubit technology.  QuTech is working on a project called Quantum Inspire and we expect them to announce in 2020 cloud availability of a small quantum machine based on their spin qubit technology.
  • Many other players are working on hardware technologies, but we are not certain how many will be announced in 2020.  These include Alibaba (superconducting), Atom Computing (neutral atoms), ColdQuanta (cold atoms), Eeroq (electronics on helium), PsiQuantum (photonic), Intel (spin qubits), IQM (superconducting), Microsoft (topological), Silicon Quantum Computing (spin qubits) and several others.  Although we believe most of these efforts will still be in development in 2020, we would not be surprised if one or two announce public availability before the year is out.

Cloud Services

The end of 2019 saw significant announcements from Amazon Web Services (AWS) and Microsoft Azure to provide cloud services for multiple different hardware platforms. We expect to see several additional cloud services announcements in 2020 and we expect to see more of these multi-platform arrangements where a cloud service will support several different hardware platforms including:

  • We expect Google to take steps in 2020 to make their machine more available on the cloud.  Google has their own Google Cloud Services group so we expect them to leverage those capabilities and compete with Microsoft’s Azure and Amazon AWS Braket services.  In addition, we note that AQT has also announced it is compatible with Google’s Cirq software so this might provide an alternate platform for Google to support in a cloud platform, if they want.
  • Although Amazon AWS’ initial Braket announcement indicated support for the D-Wave, IonQ, and Rigetti platforms, one additional statement included in the announcement is that they are expecting to announce additional partners soon.  Their strategy is to develop cross-platform developer tools so that an end user can program using Amazon’s front end can switch between hardware platforms relatively easy. How they intend on doing this will be interesting because each of those platforms are quite different.  There are still a lot of details not yet available about AWS’ Braket offering including pricing, software, and additional details on IonQ’s hardware.  We expect to hear more about these in 2020.
  • Microsoft also announced that their Azure cloud platform will partner with Honeywell, IonQ, and QCI.  Their software front end will be the established Q# programming language and the Quantum Development Kit. Like the AWS announcement a lot of details are still not available including pricing and details on the hardware and we expect to hear more in 2020.  Microsoft also announced that they will eventually hook up their topological based quantum computer to the Azure platform, but we think achieving this in 2020 may be a stretch.
  • IBM has been the most aggressive cloud platform vendor in establishing a dedicated quantum data center in Poughkeepsie, New York. As of September they had 10 machines quantum machines available on the cloud and were in the process of adding four more.  And this does not include their recent announcements to install additional IBM Q System One machines in both Germany and Japan nor does it include any machines that are using for internal research and development.  IBM may also opt to provide support for an alternate hardware platform to provide their users a means to compare different technologies.  IBM recently announced that their QISKIT software platform now also supports AQT’s ion trap technology and that it only took one week’s work to implement this.  We expect that other hardware technologies could be easily supported in QISKIT if IBM chooses to do so. There is also an open source module called Forest backend for QISKIT, but we are not sure if IBM is ready to support another superconducting quantum computer.
  • We may see in 2020 even more cloud vendors jump in to provide some form of quantum cloud service. This could include both classical cloud vendors who don’t want AWS and Azure to get too far ahead of them as well as some of the emerging hardware companies that desire to set up their own cloud service. Also, some application software companies that want to act as resellers and bundle their own application level software with one of the hardware platforms and market both together as a package.

Application Software

In 2019 we saw a bunch of new software startups that are focusing on specific applications and offering their services to end users and we expect this trend to continue in 2020.  The focus so far has been working on proof-of-concept applications so that the end users, as well as the startup quantum software companies, can identify specific real world problems and solutions that will prove to be commercially useful when the larger machines are available. We do expect to see significant development in 2020 in new application libraries, continued enhancements to many of the open source development platforms and improved performance in simulators. For the gate-based machines, we do not expect to see more than a handful of applications, at best, start being used in a production mode in 2020.  In fact, some quantum researchers are not sure there will any significant number of production applications in the NISQ era. They argue that this will not occur until the larger error corrected machines are available later this decade.

D-Wave has been working with end users for several years now and over 200 early applications have been developed for their quantum annealing machine.  Many of these are proof-of-concept applications that aren’t meant for daily production such as the recent Volkswagen experiment to optimize bus route optimization during the recent Web Summit conference in Lisbon.  Because D-Wave started earlier, has a more focused set of potential applications and a larger number of available qubits (even though the qubit quality levels may not be as high as the gate level machines), we do expect in 2020 to see a handful of these early applications being using on a production basis for commercial use.   So on the very important measure of using a quantum computer for commercial use we do expect that D-Wave to beat out the gate level machines, at least in 2020.

Optimizing Compilers (aka Transpilers) and Qubit Control Firmware

Perhaps not as widely appreciated is the importance of the backend compiler and qubit control firmware that translates the application program that user might develop to the specific electronic signals that control the operation of the qubits. The goal is to execute the user’s program in a way that minimizes qubit and gate count, minimizes circuit depth, and provides the best accuracy for the overall solution.  This is accomplished by rearranging the gates that an end user may initially input into something equivalent, but more efficient, as well as optimizing the microwave or laser pulses that control the individual qubits.

This software can become enormously complex but is critical to optimizing the performance of the hardware. Google has indicated that they would have been challenged to successfully complete their Quantum Supremacy experience without the use of this software. Leaders in this area include Q-CTRL and Quantum Benchmark, and we have also heard good words about Cambridge Quantum Computing’s (CQC) optimizing compiler for the IBM machines which they claim to be superior to the ones provided by IBM in QISKIT.

We expect to see significant developments in this area in 2020.  In December, IBM made one of their machines available for external researcher to experiment with pulse level controls of the hardware.  Our belief is that a lot of work still needs to be done in this area with a lot of opportunity to make great strides in this area in 2020.

Wish List

There a few things we would like to see happen in 2020 that would benefit the QC industry overall, but we are not sure how much progress will be made on these in the coming year. Nonetheless, we will list them here to help promote advancements in these areas.

  • Quantum computing nomenclature should be standardized so everyone uses the same terminology.  Unfortunately, we see differences in how the same thing is described by different people and this can be quite confusing to a newcomer. (For example, try viewing a collection of different images of a Bloch sphere and you will notice that some versions have switched the positions of the X and Y axes.)
  • We would like to see a standardized hardware agnostic programming language for gate level machines with an associated requirement that all higher level software platforms be able to both import and export programs to and from this language.  This would encourage interoperability and make it much easier to convert a program from one platform to another.
  • With all the new machines coming on-line in 2020 there will be a lot questions about performance benchmarking to compare the different machines.  Although IBM has developed their Quantum Volume metric, it is not clear how widely this is supported within the rest of the industry.  We’d like to see a standardized benchmark, analogous to LINPACK for supercomputers, that everyone can agree upon to characterize the performance of the quantum computers.

Non-Technical Factors

There are other non-technical factors that we expect to hear more about in 2020.  There is a continuing concern that we are not developing a quantum trained workforce as fast as we should. Steps are being taken to improve education and training programs but it is not clear that these are enough. Another factor we see is the growing tension between researchers and government officials over export and visa controls. The researchers are concerned with hampering overall progress in QC research if information cannot flow freely, while the government officials are concerned with maintaining their country’s control over the technology so that it is not use unfairly by a country’s adversary.

Finally, there is always talk of a quantum winter.  We don’t see that in 2020 as many of the government funding programs that we have reported on will start to kick in.  In addition, there will be continued investment in the private sector, particularly from the large classical computing companies that don’t want to miss out. On a positive note, a few of the smaller software startups have already told us they expect to be profitable in 2020.


So we expect that 2020 will be another exciting year for the quantum community with a lot of progress being made. Many of the developments will be extrapolations of things we saw in 2019, but we also expect a few surprises where new technologies or new players come in and provide something new that we did not expect. Still the developments in 2020 will represent continued progress in a technology that will require several decades to reach its full potential. 

We wish everyone working in this area the best of success and look forward to reporting on your developments as the year progresses.

December 30, 2019