In the 1980s, two scientists theorized that if a quantum computer could be designed, it would be able to simulate and model problems that traditional systems could not.
In 2019, we started tracking quantum computing – specifically, IBM’s design and development efforts to make quantum computing a reality. In one of our first reports, “Quantum Computing: Measuring the Momentum,” we stated that: “The roll-out of quantum solutions is being throttled by quantum computing capacity. In other words, the types of problems that quantum computing can solve are already well-known – but much more computing capacity (and better noise reduction combined with fault tolerance and better qubits) are needed to tackle larger problems. In short, quantum computing at present is hardware constrained.”
IBM quantum computers operated at 53 qubits way back then…
Fast forward to today. In November 2021, IBM introduced a new quantum processor – the 127-quantum bit (qubit) ‘Eagle’ processor. My, how things have changed…
And this story keeps getting better and better. IBM recently produced a roadmap video that describes how and when it plans to deliver a processor capable of 4,158+ qubits (target: 2025)!
In quantum computing, IBM states that its goals are to:
- Focus on building lots of quantum circuits on high-quality quantum hardware;
- Regularly increasing the performance of quantum processors;
- Continue to find ways to deal with errors; and,
- Simplify how quantum computers are programmed.
As IBM seeks to meet these goals, it will focus on scale, quality and speed challenges.
The hardware roadmap
As IBM closes out 2022, it will have delivered improvements in:
- Dynamic circuits;
- 433 qubit “Osprey” processor next step in scale;
- Quantum Volume; and in
- Speed: 1.4K CLOPS to 10k CLOPS.
IBM’s quantum process roadmap over the succeeding three years looks like this:
- 2023 — multi-chip quantum processors with newly designed chip-to-chip couplers;
- 2024 — quantum parallelization of quantum processors (1,386+ qubits); and,
- 2025 — quantum parallelization of multi-chip quantum processors (4,158+ qubits – the Kookaburra project.
Concurrent with speed improvements, we also expect to see improvements in error mitigation and error correction – and at some point, IBM will be using “circuit knitting” (running a problem across multiple processors and knitting the solution back together when the problems are solved).
In 2024, expect to see IBM’s codename “Heron” processor, which will exploit new gates, new tunable couplers, and simultaneously limited crosstalk to improve quality (this is a complete redesign). IBM will also be working toward trying to control multiple Herons with the same control hardware using a new device that allows quantum gates between quantum processors (a chip-to-chip coupler). It will yield results similar to using a single large processor, only the work will be processed across multiple quantum processors.
Heron will be followed by the introduction of a planned long-range coupler that will enable 1,386+ qubit systems (Flamingo).
For developers, over the next several years, IBM will focus on simplifying programming for kernel, algorithm and model developers:
Kernel developers focus on making quantum circuits run better and faster – they focus on delivering dynamic circuits. For kernel developers, IBM will extend what the hardware can do (reduce circuit depth, allow alternative models for algorithms and perform parity checks for QEC).
The algorithm developer uses the kernel-created circuits to run classical routines to create applications that demonstrate quantum advantage
Model developer uses the applications to find solutions to complex problems in their specific domains. For domain experts to model solutions, application services will help integrate machine learning and kernel algorithms. Expect IBM partners to exploit these in new application solution development.
Expect hardware and software extensions for each of the above classes or quantum developers. For instance, in 2021, the Qiskit runtime will be further improved. New primitives will be the bedrock of unique user programming experiences in the future. Examples: new primitives such as Sample (for search applications); and Estimator (for, just like the name says: estimating).
In 2023, expect threaded runtimes (these will enable Qiskit to act as an API).
As for error correction and error mitigation, look for probabilistic error correction.
Next year IBM plans to deliver “Quantum Serverless,” bringing new functionality into the stack (with circuit knitting – described earlier).
On the tool’s front, expect improvements in the Qiskit runtime and new functionality such as entanglement forging, circuit knitting toolbox, and threaded runtimes.
Over time, expect classical circuits with real-time quantum computing (known as “Dynamic Circuits”).
IBM will also mitigate and correct errors with near-time classical computing (Qiskit runtime).
Elastic classical computing will be used to knit quantum programs together to solve significant problems (quantum serverless). And the blending of classic computers with quantum computers will help create quantum-centric supercomputers.
Quantum Volume is the fundamental performance metric that measures progress in the pursuit of Quantum Advantage, the point at which quantum applications deliver a significant practical benefit beyond what classical computers alone are capable of. Although substantial improvements in qubit capacity have been made since we wrote our first review of quantum computing back in 2019, Quantum Volume is still the limiting factor of quantum computing, not a lack of use cases.
Today’s quantum computers can already solve a wide array of computational problems. But when IBM reaches its goal of 4,158+ qubits (target: 2025), expect to see two things. First, IBM will have worked hard to blend classical computers with quantum computing. Integration will vastly improve. Second, when combined with other new technologies (such as IBM’s new couplers and control devices) and with continued infrastructure/program advancements, the day is rapidly arriving where IBM (and its partners) will be able to reliably tackle highly complex problems in advances in chemical simulation, scenario simulation, optimization, and artificial intelligence/machine learning (AI/ML), and in chemicals and petroleum (advances in oil shipping/trucking, refining processes, feedstock to product and drilling locations). Also, expect advances in distribution and logistics which will include freight forecasting, irregular behaviors (Ops), disruption management, and distribution supply chain improvements. In financial services, quantum computers will bring advances in derivatives pricing, irregular behavior analysis (fraud analysis) and investment risk analysis. In healthcare and life sciences, advances can be expected through accelerated diagnosis, clinical trial enhancements, genomic analysis, and medical/drug supply chain activities. Finally, quantum computing will aid in quality control, process planning, manufacturing supply chain activities and fabrication optimization in manufacturing.
Also, don’t forget, there will be big advances in programming quantum computers – and an expanded ecosystem.
The blending of classical computing with quantum computing will be something to behold.
For an in-depth look at IBM’s quantum processor development roadmap, you’ve got to see this video.