At its annual IBM Quantum Summit today, IBM made several announcements of new advancements in its quantum technology and their quantum network. Some of these had been previously mentioned in roadmaps, technical papers, press releases and other venues, but more details on these have been provided today.

IBM 433 Qubit Osprey Processor

IBM first mentioned their Osprey processor on their long term roadmap in September 2020 and now it is being announced per their original schedule. A key purpose of the machine is to introduce several new technology required to scale up the number of qubits. A technology being introduced with this generation include cryoflex cabling for carrying the qubit control signals in a more efficient manner. IBM indicates that this new cabling will provide a 70% increase in wiring density and a 5x reduction in price per line. They have also developed a new generation of smaller and faster control electronics using FPGA to handle the larger number of qubits as well providing better support for dynamic circuits and improve the speed for better CLOPS. The chip is physically quite large as shown in the pictures below and it continues to use the 3D packaging technology and other innovations that IBM introduced with the 127 qubit Eagle (ibm_washington) device.

Picture Showing the Relative Sizes of Several Generations of IBM Quantum Processors. Credit: IBM

Picture Shown Exploded View of Osprey Processor Showing 3D Packaging. Credit IBM.

Previous Generation Coaxial Cable Wiring (Left) and New Cryoflex Cabling (Right). Credit: IBM

IBM’s focus in this development was to introduce key technologies needed to scale up the number of qubits. However, many elements of the qubit architecture are similar to the ones used in the Eagle (ibm_washington) device and it will continue to use IBM’s Quantum System One packaging introduced in 2019.  Osprey will be made available to members of the IBM Quantum Network in Q1’2023. The current version of the device, version R1, has many of the qubit quality metrics similar to previous generations with T1 coherence times in the 70-100 microsecond range. The next revision of Osprey, dubbed R2, will incorporate several improvements in coherence times that were first tested on one of the 27 qubit Falcon devices and will improve this coherence by a factor of 2-3X. However, we do not expect to see drastic improvements in the gate fidelities from the previous generations nor do we expect the Osprey processor to show higher levels in IBM’s Quantum Volume metric. Improving these will be a key focus in IBM’s next generation Heron processor which is slated for introduction in 2023. 

New Quantum Volume and CLOPS Measures Achieved

For some of its developments, IBM will use a few of the previous generations to test innovations and improvements and this has been the case with new levels announced today in the Quantum Volume and CLOPS (Circuit Level Operations per Second). They recently have been using their 27 Qubit Falcon device to test these improvements and have now reached a Quantum Volume metric of 512 and a CLOPS measurement of 15.7K. The CLOPS improvement is quite significant because they started the year at 1.4K, showing a 10x improvement this year.

Error Suppression and Mitigation

As we have mentioned many times, one of the limiters to the usefulness of quantum computers is the level of errors in current generation of machines. IBM has introduced certain techniques reduce these. Error suppression uses a technique called dynamic decoupling which modulates the control pulses in a way that will reduce decoherence and crosstalk. Error mitigation uses a variety of different algorithms to provide more accurate answers using different techniques. Two algorithms that IBM is using are called Zero Noise Extrapolation and Probabilistic Error Cancellation. Additional information about these algorithms is available in two different blog articles published by IBM that can be seen here and here.

The tradeoff of using these algorithms is that they can substantially increase the overall runtime of a job because they could require multiple iterations to before providing the answer. So, the user is forced to choose between trading off accuracy versus runtime. In order to make this simpler for end users, IBM is implement a simple control that will allow end users to specify how much mitigation they want and then have the software implement it automatically. The control is called the Resilience Level and can be set anywhere from 0 – No Mitigation Algorithms up to 3 – Maximum Mitigation Algorithms, with levels 1 and 2 somewhere in-between.

Dynamic Circuits

IBM disclosed their intention to make available dynamic circuits, sometimes called mid-circuit measurement, at their 2022 IBM Quantum Summit  and it is now available on 18 systems. This capability allows an end user to measure a qubits and take a branch in their code depending upon the outcome of the measurement. The qubit that was measured can then be reset and reused later in the program. This capability can reduce circuit depth and/or reduce the number of qubits needed to implement an algorithm.  Although we have not seen too many user programs take advantage of this capability, it is still pretty new. It is expected that this capability will become even more important as error correction techniques are implemented in order to detect and then subsequently correct qubit errors. This feature is now supported in OpenQasm3 as well as Qiskit. IBM has publshed a blog that describes this capability in more detail which you can find here.

Circuit Knitting and Quantum Serverless

IBM also announced it has made alpha releases of two capabilities disclosed at the IBM Quantum Summit in 2022. Circuit Knitting entails splitting up a large quantum program into separate pieces that can be run independently. The outputs of the runs can be then fed into a classical computer and then combined to provide an answer. This approach may allow a user to run a program larger than could fit into one machine or allow a program to run faster if the separate pieces can be run in parallel on multiple quantum processors. IBM also announced a year ago to implement serverless computing which helps a user to run hybrid classical/quantum programs without the user manually have to specify how to configure the systems or orchestrate this. Serverless computing can handle this automatically and make it easier on the user.

Continued Progress on Developing the IBM Quantum System Two Packaging

Another development that IBM first disclosed a year ago was new packaging called the IBM Quantum System Two. This system is designed with modularity in mind for better scalability, with the vision of entering the era of quantum advantage. The design will accommodate a variety of different configurations as well as a means to facilitate networking by enabling multiple systems to be placed physically close together.

Top Down Diagram of Possible IBM Quantum System Two Configurations with 1 (Left), 2 (Middle), and 3 (Right) Cryostats

In the picture above the cryostats (C1, C2, and C3) are the hexagonal shapes with the control electronics placed in the wings attached to the cryostat. The wings are flexible and extendable to accommodate the required electronics for different numbers of qubits. And as shown in the middle and right pictures, the cryostats can be placed together to minimize the distance for communication between qubits in different cryostats. With the Kookaburra chip each cryostat could hold 4,158 qubits so the configuration with three of them could hold 12,474 qubits or more.

The IBM Quantum System 2 packaging is scheduled for late 2023 and is expected to be officially announced at the IBM Quantum Summit 2023.

Continued Expansion of the IBM Quantum Network

IBM announced the continued expansion of the IBM Quantum Network reaching over 200 members. They also announced nine new quantum innovation centers to reach a total of 34. In addition, they listed 19 industrial customers working on use cases in simulating nature, mathematics and processing data with complex structure, and search and optimization.

Quantum Safe Services

Now that NIST has selected it first set of post quantum algorithms, IBM will be placing increased emphasis on helping with clients become quantum safe. As we reported a few weeks ago, IBM is now working with Vodaphone and the GSMA on a taskforce to guide mobile cellular network operators to become quantum safe. The 2022 IBM Quantum Summit was the first quantum summit that covers this topic and we expect IBM to market their quantum safe and quantum communications capabilities together more often in the future. It is quite likely that early adopters of quantum computers will also be early adopters of quantum safe communications in the future, so this makes sense.

IBM’s 100 Qubit, 100 Level Challenge by the End of 2024

Finally, IBM announced a challenging goal that they have set for themselves. They want to build a tool, which is capable of estimating noise-free observables of circuits consisting of 100 qubits and a depth of 100 within a day’s runtime. It will require using their Heron processor being developed for next year with an error rate of less than 0.1%. The Heron will incorporate additional features, such as tunable couplers between qubits to reduce crosstalk so it can reach these improved error rates. We would expect that the Quantum Volume measurement on such a device will be much improved.


IBM is continuing to make progress in its quantum efforts and are making significant efforts across a variety of fronts with a goal of establishing quantum-centric supercomputing a reality within the next few years. Although they are expected to offer error corrected quantum machines over the long term, they are of the belief they will be able help customers achieve quantum advantage with the improved NISQ level processors they are planning to offer.

One area where IBM is to be commended is in their transparency. In a time where there is a lot of hype about quantum, many end users are uncertain about what to believe or what to plan for. By providing clear roadmaps, and just as important, meeting the milestones on their roadmap, they are reassuring end users about the viability of quantum technology and giving them confidence about making an investment for incorporating quantum within their organization.

We are hoping that other providers will follow suit and take a similar tack. It doesn’t require a multi-million dollar budget or hundreds of engineers for a provider to be more transparent. This will allow end users to make a more informed choice on how to get involved with quantum technology and discourage them from holding off because they don’t know what’s up ahead.

IBM’s press release summarizing these announcements can be accessed here and an associated blog article that describes how these technologies fit into their vision of Quantum-Centric Supercomputing is available here. Also, IBM has posted on Flickr several videos and photos related to their announcement that you can see here.

November 9, 2022