IBM touts ‘Quantum Serverless’ as it eyes path to 4,000-plus qubits

news image

We are excited to bring Transform 2022 back in-person July 19 and virtually July 20 – 28. Join AI and data leaders for insightful talks and exciting networking opportunities. Register today!

IBM today updated its existing quantum computing roadmap, foreseeing a 4,000-plus qubit machine by 2025. Included in its plan is orchestration for more potent combinations of classical and quantum resources. 

Such combinations, which comprise what may be called “quantum-centric supercomputing,” will place the new quantum crew more firmly in the world of high-performance computing, where technologists tackle some of humankind’s most pressing challenges. 

IBM said it expects to deliver in 2023 software tooling that allows developers to run jobs that help them tap into both quantum and classical computing resources. IBM said it sees no letup in the steady advance of quantum scaling.

The company uses the term “Serverless Quantum” to describe an architecture that deploys infrastructure that assigns work to processors of either the classical or quantum variety.

“Serverless Quantum is part of getting the world ready for quantum-centric supercomputing,” according to Jerry Chow, IBM fellow and director of quantum infrastructure. He likened the effect to today’s elastic computing methods, “in the sense that you have computing resources on demand.”

Speaking with VentureBeat, Chow said the idea is that parts of problems can be handled by either classical CPUs, GPUs or non-classical QPUs, or quantum processing units. This is enabled by a common service layer that efficiently orchestrates, or arranges, workload assignments. 

Better hardware and software methods for interchip communications is also important in this quantum quest, Chow indicated as he discussed future plans for quantum “circuit knitting.” 

“That’s where you can take larger problems,” he said, “and find ways to break them down into small-order quantum circuits, and knit or take all the results and put them back together.”

The problems imagined are manifold, and some appear especially pressing. 

Simulating nature with quantum systems, for example, is seen as a path to efficiencies that could drive substantial environmental, social and governance improvements. 

Also, world governments are now seen in a race to succeed in quantum computing, as quantum systems could crack long standing cryptographic coding practices. Viewers suggest adversary governments are gathering encrypted data now, to process when quantum breakthroughs occur. 

Sticking to their circuit knitting

IBM has indicated Serverless Quantum capabilities in its Qiskit Runtime software, including circuit knitting, will support distribution of parallelizable work in 2023. As well, improvements in error mitigation and suppression are expected to improve quality of quantum results, the company said. These are due for Qisket as part of the roadmap in 2024 and 2025.

The bird is the word when it comes to code names for IBM quantum processors. Last year, IBM released a 127-qubit Eagle processor to be followed this year by Osprey, a 433-qubit unit. An increasingly powerful flock of IBM qubit processors will follow. IBM’s roadmap anticipates: a 133-qubit Heron processor to appear in 2023; a 462-qubit Flamingo processor in 2024; and the 4,158-qubit Kookaburra device in 2025. 

While Heron’s qubit count may look like a step back, it’s important to IBM’s overall plans to knit clusters of qubit processors together. That is vital to IBM’s assertions of continued scalability. 

“What’s exciting about Heron is that it is defined in such a way that they’re going to be classically controllable [via] the same control hardware infrastructure. So what that means is effectively, we’re going to have quantum computing with classical communication and classical parallelization,” said Chow, who is expected to further detail IBM’s  quantum roadmap this week at the Inside Quantum Technology Conference 2022 in San Diego, Calif. 

Quantum timelines

Timelines for quantum computing have been closely watched since qubits first emerged from the lab as part of physics experiments. A common joke holds that today, just as at the turn of the century, “quantum computing is five to ten years away.” 

But IBM’s timelines for quantum computing have usually been realistic, according to Brian Hopkins, analyst, Forrester Research.

“IBM has the most detailed roadmap of any of the full stack quantum computing vendors I cover,” he told VentureBeat via email. “Furthermore, they are establishing credibility with the market by completing roadmap milestones per their plan.”

That gives Forrester’s Hopkins confidence in IBM’s 4,000-plus qubit projections, although he emphasizes that IBM is still refraining from any meaningful prediction on when its computers and software will be able to demonstrate advantage – in the form of a positive ROI — for any problem domain. 

Hopkins goes on to say this is valuable, in that the company will have greater credibility when it does prove such an advantage.

“Even when we hit advantage for one problem area, it will still be many years before quantum computers are scalable and powerful enough to be applicable to the many areas where we see potential,” he said.

We love qubits – but….

Skepticism of quantum computing timelines is not likely to abate, but elements of IBM’s new timeline suggest the technology may be moving forward faster than some skeptics thought.  

While “qubit counts” will get the headlines here and elsewhere, the technology advances that surround qubit research bear closer watching, as do hybrids that match classical and quantum computing to different parts of problems.

“We love qubits – but the exact number is not as important as [the fact] that we have defined the path for the technology to keep scaling,” said Chow.

VentureBeat’s mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Learn more about membership.

Read More