IBM Corp. today announced a new roadmap that it says will usher in a new era of large-scale and practical quantum computing.

The plan involves creating new modular architectures and networking systems to support more powerful quantum processors with higher qubit counts. At the same time, the company will design a more intelligent software layer to handle quantum workloads more efficiently.

There’s every reason to believe IBM can achieve its ambitious targets. After all, the company delivered on all of the targets on its original quantum roadmap, including the most recent milestone of a 127-qubit processor dubbed “Eagle” that it unveiled last year.

Eagle’s architecture laid the groundwork for new quantum processors with vastly improved qubit counts. The company has also improved processor runtimes by an impressive 120 times through the Qiskit Runtime software platform it also announced last year.

By the end of 2022, the updated roadmap calls for IBM to launch a new 433-qubit chip called “Osprey” and make substantial progress on its goal of building a more frictionless development experience with Qiskit Runtime and workflows built into the cloud. Then by 2025, the company should be ready to launch “Condor,” its first chip with more than 1,000 qubits.

IBM explained that it’s employing a modular approach to try to scale up its quantum processors. This will involve developments in three key areas, with the first being new capabilities that can classically communicate and parallelize operations across multiple processors.

The company said that will clear the way to a broader set of techniques necessary for practical quantum systems. Those include improved error mitigation techniques and intelligent workload orchestration, accomplished by combining classical compute resources with expandable quantum processors.

The other techniques include the deployment of “short-range, chip-level couplers” that will closely connect multiple quantum chips together to form a single but much larger processor — a kind of modular approach to building larger chips, if you will. Finally, IBM will work to develop what it calls “quantum communication links” between these modular processors, so it can link clusters of chips in such a way that it creates a much larger system.

IBM said it hopes to combine these techniques and hit its goal of a 4,000-plus-qubit chip called “Kookaburra” by 2025.

Of course, IBM understands that practical quantum computing requires more than just the incredibly advanced hardware. To that end, it will simultaneously work to achieve targets around quantum error suppression and mitigation.

That’s where IBM Qiskit Runtime comes in. The software works to minimize the effects of noise on quantum applications while encapsulating common quantum hardware queries used in algorithms within an easy-to-use interface. IBM plans to expand Qiskit Runtime’s capabilities and allow developers to run algorithms on parallelized quantum processors to speed up applications

In 2023, it will also add Quantum Serverless capabilities to its core quantum computing software stack. That will allow developers to trade off and switch between classical and quantum computing resources as required.

Dario Gil, IBM’s senior vice president and director of research, said the company’s initial successes means it has clear visibility about what’s required to deliver practical quantum computing.

“With the quantum serverless operations and the advances in hardware, software and theory goals outlined in our roadmap, we intend to usher in an era of quantum-centric supercomputers that will open up large and powerful computational spaces for our partners and clients,” Gil said.

Images: IBM