What is one to make of the quantum computing market? Energized (lots of funding) but still chaotic and advancing in unpredictable ways (e.g. competing qubit technologies), the quantum computing landscape is transforming at a blurring rate. To mix metaphors, think of the quantum computing technology community as being in its own state of superposition and not really ready to coalesce into a coherent picture.
Bob Sorensen, chief quantum computing analyst, Hyperion Research, took a creditable stab at describing quantum computing sector at last week’s virtual HPC User Forum. There’s already QC market, albeit small now (~$320M in 2020) but which is projected to grow to $830M in 2024 (27 percent CAGR) according to Hyperion.
Sorensen focused broadly on market dynamics and steered clear of deep technology dives. A fair amount of the material was familiar from the spring HPCUF but there were updates too. He set the tone with this early comment on competing qubit technologies (semiconductor, trapped ions, optical, topological, etc.), “I like to say, perhaps somewhat provocatively, [that] if we could move the clock forward, say 15 years, the quantum modality of choice may not yet be on this list.”
Many observers would agree. Quantum information sciences R&D and its commercialization efforts are expanding so quickly that it’s difficult to say much about what the final quantum computing systems or application landscape will look like. What does seem clear is that there will indeed be a quantum sector of some sort. Sorensen presented charts, packed tightly with QC participants (government, corporate, regional) and QC technologies to emphasize the scope of the current frenzy if not the details.
“If there’s one takeaway that I’d like everyone to have from today’s discussion, it’s that quantum computing is not a standalone technology. It’s not ‘here’s classical computing on one island, and here’s quantum computing on another island.’ Quantum computing is more about another tool in the toolkit for advanced computing. Think of it as a version of a GPU in the sense that quantum computing offers some significant performance advantage for a key but narrow set of applications that ultimately complements the overall advanced computing architecture but will not supplant it.”
Let’s start with the global nature of the market and dominant players:
Citing the number of papers published by country of origin, Sorensen said, “This is not something that the U.S. owns by default, based on 40 years of its leadership. I did a very simple query (of quantum-related paper publications) about two months ago, and ended up receiving about 17,000, high quality R&D publications in the last five years. What I find interesting is the number of other countries in the world that are posting reasonable amounts of research activity in these publications. This is an international phenomenon.”
“You see countries like India and Canada and the U.K., [that are] not exactly IT powerhouses but are playing quite aggressively to get involved early on in the quantum computing sector,” Sorensen said.
He cautioned against assuming China is leading in research based on its showing on journals: “The reason the Chinese academies show up at the top is because they’re responsible for the preponderance of what’s going on in China. If you want to know what’s going on in China, you basically look at these two organizations as kind of a harbinger of what the rest of the nation is doing. Whereas in the U.S., which had about the same number of R&D publications, it’s a much broader base of more research facilities, universities, and government research facilities.”
Sorensen cited the growth of non-competitive collaborative efforts going on around the world. One example is the Quantum Economic Development Consortium in the U.S., which is funded by the U.S. government. “It is still a very free-flowing, open organization to foster the non-competitive aspects of quantum computing,” said Sorensen. We’re seeing the same things with a quantum industry consortium in Europe, where we have a mix of QC suppliers and major industrial players.”
In Japan, he said, telecommunications giant NTT has initiated a cooperative organization of 11 leading Japanese industrial players.
“We saw the same thing in the last few months in Germany [where] ten industrial players – including automakers, advanced manufacturing, and investment houses – are involved in thinking about how use cases can drive what we see in terms of quantum computing offerings from the suppliers,” said Sorensen.
Virtually all of these efforts are intended to nourish regional and national infrastructures for commercial quantum computing. On their agenda are quantum tech standards, supply-chain frameworks, financing, use case identification and go-to-market strategies. Their rather recent formation reflects the growing worldwide expectation that quantum computing and its related quantum technologies will make the leap to commercialization soon, though no one seems to have a firm handle on when that will be.
“The timeframes are unpredictable at this point. The sector makes advances in leaps and bounds, or then it can sit still for a little while,” said Sorensen
Hyperion’s research, said Sorensen, suggests potential quantum computing users are willing to make the switch to quantum technology for about a four-to-five-year jump over that normal HPC advances would offer. He said that’s about 50x improvement over current methods. “[They] are saying, “All I need is a four-to-five-year boost over my competitors, based on what I would see in the classical world. You give me a 50x performance improvement, [and] that’s really what you’re going to get in classical HPC if you wait five years [for HPC advances]. [This is an] important point to make to suppliers in the quantum sector, that users want competitive advantage and economic payoff and aren’t looking for [too] aggressive, far reaching potentials (such as quantum supremacy),” Sorensen said.
So how does the quantum computing community get from here to there?
Quantum technology is not ready now for prime time agree most observers, although there are many “pilot/proto” applications and use cases being explored. One promising near-term trend is development of quantum-informed and quantum-inspired approaches. Zapata Computing, an example of the former, uses quantum computers to generate input (mostly random numbers) for use by classical systems. There are also a number of quantum-inspired optimization techniques being run on classical systems; these are quantum computing algorithms that have been adapted for use on classical digital systems. Of course, the pure-play quantum companies like Rigetti, IBM, and D-Wave all have vigorous commercial engagements at various stages of development.
Most of these are still in POC stage projects. Still, one has the sense that there are so many individuals, companies, academies, and governments frantically toiling away at quantum computing that accelerated progress is inevitable. We’ll see.
Sorensen cited the usual applications. He noted much of the attention has been around physical simulation (chemistry, materials science) where quantum computing’s inherent advantage of using superposition and entanglement should pay dividends. Optimization is another big class of problems being tackled and perhaps closer to near-term practical use.
“There’s also some interesting work going on in quantum accelerated machine learning capabilities to really help augment what we see going on in the classical world of machine learning, deep learning and other variants of artificial intelligence,” said Sorensen. “One of the most popular [right now] is the optimization opportunities in quantum – the ability to take existing algorithms or objective functions to figure out better ways to maximize an outcome, for example, figure out the best way to load a large group of luggage and other materials into the cargo hold of an Airbus airplane as it turns around at the gate,” he said.
(Update: The table (QC Software Suppliers Abound Across QC Modalities) below incorrectly shows Agnostiq as only supporting IBM. The company reports its “software actually supports all of the following vendors: Amazon Braket; Google; IonQ; Xanadu; IBM; D-Wave; Toshiba; [and] Rigetti.” The table was reproduced from the Hyperion Research presentation. We’ve kept it in the article because, despite the error, it provides a sense of how large the community of quantum software companies is becoming.)
All summed up, Sorensen said, “One of the interesting things that makes the sector so amazing and confusing and interesting to watch is development is happening in many directions and [in] parallel,” said Sorensen. “[W]e’re seeing different hardware qubit modalities. People are thinking about interconnects. If we build 1000-qubit processors, how can we configure that to a million-processor-system? There’s the idea of quantum LANs. How do we think about the overall architecture of QC to really start to field systems that can actually deliver impressive performance in an industrial [setting].”
Near-term, many challenges remain for the quantum computing sector. Not only the technology itself, but workforce issues and business models. Stay tuned.
Slides from Bob Sorensen’s presentation at the fall 2021 HPC User Forum