Kit Wai Chan - Fotolia
The upcoming system, scheduled for release in late 2023, will mark the debut of the company's Condor processor. IBM believes Condor will allow developers to jump over many technical hurdles preventing them from solving a range of complex scientific and business problems.
"We think Condor represents an inflection point that marks our ability to implement error correction and scale up our devices, but yet still be complex enough to explore solutions for problems that can't be solved by the fastest supercomputers," said Jay Gambetta, IBM Fellow and vice president of quantum computing, at Wednesday's IBM Quantum Summit.
More planned for IBM
Before Condor arrives in late 2023, IBM will release three other lower-scale systems.
Very quietly earlier this month, IBM made its 65-qubit Hummingbird processor available to developers as part of its IBM Q Network. The system features an 8:1 readout multiplexing, meaning it can combine readout signals from eight different qubits into one. This reduces the amount of wiring and components needed for readouts and improves the system's ability to scale, according to IBM.
Next up will be IBM Quantum Eagle, a 127-qubit system scheduled for delivery in 2022. The system's processor will be the first to use concurrent, real-time classical compute capabilities, making it possible to execute a much broader array of circuits and codes.
Bob SorensenSenior vice president of research and chief analyst, Hyperion Research LLC
Also in 2022, IBM will deliver the 433-qubit Quantum Osprey system. It will feature more efficient and denser controls and cryogenic infrastructure to better ensure scaling of the processor without sacrificing the performance of individual qubits.
More important than the boost to the upcoming systems' raw processing power are the planned improvements to underlying technologies such as the control systems and communication capabilities.
"They aren't moving towards product development yet, but they are moving in the direction of looking at things more as engineering problems as opposed to basic science problems," said Bob Sorensen, senior vice president of research and chief analyst for quantum computing at Hyperion Research LLC. "They are thinking about things from a productization perspective, like how people can talk to quantum systems in a realistic way when they start to scale the systems up."
Sorensen added that the significant scale-up in qubits with a quantum computer such as Condor adds critical value in terms of introducing new capabilities. When systems have 1,000 qubits or more, the possibility of introducing more effective error correction capabilities can be introduced, which is critical to improving quantum systems reliability in production environments.
"Right now there aren't enough qubits to do any realistic kind of error correcting," Sorenson said. "But with 1,000-plus [qubits], you can start down that road."
Issues that may hinder scale-out
While delivering a quantum computer with 1,121 qubits is impressive, another analyst said IBM will have to solve a number of other technical issues before users can apply the added processing power to practical solutions.
"It's going to be interesting to see how they scale this superconducting architecture to 1,000 qubits, because right now each qubit needs three or four coaxial cables for controlling readouts," said James Sanders, a research analyst on the cloud and managed services transformation team at 451 Research. "That means you need 3,000 to 4,000 cables, so the way they approach scaling the orchestration layer is going to be important for them to be able to do this."
The 12-foot-tall cooling system IBM is developing, codenamed Goldeneye, is being designed primarily for the 1,121-qubit Condor system. Gambetta said IBM has begun feasibility tests on the unit. He added that the company envisions a future in which quantum interconnects link dilution refrigerators, each holding 1 million qubit systems, similar to the way the intranet links supercomputer processors now. This will allow for the creation of massively parallel quantum systems.
"We see a fault-tolerant quantum computer that now feels like an achievable goal within the coming decade," he said.
The new quantum hardware roadmap resides at the heart of IBM's larger mission, Gambetta said -- namely to design a full-stack quantum computer that can be deployed in the cloud, enabling anyone anywhere in the world to develop exploitive applications.