It’s slowly emerging from research and heading toward high-volume production.
Key Takeaways:
Quantum computing has imprinted itself on our society as a weird, wacky way of computing that most of us can’t comprehend. And some of that is true, especially given the old chestnut that says, “If you think you understand quantum, then you obviously don’t.”
Is this a new computing paradigm that will overrun what we’ve been doing for decades? The answer is no, but there definitely are problems that will benefit from a commercial offering once it is available. That timeline, however, isn’t clear.
The pace of quantum development has been slow, with fussy challenges generally making it a fussy technology. Further critical development areas include the actual computing hardware (such as the qubits), how to handle error correction, and how specific algorithms will work for the kinds of problems that quantum can address. It must also mature from the domain of researchers into a full-scale commercial technology.
The state of the quantum computing industry
Quantum technology is at a stage where maximum creativity meets maximum chaos. The industry hasn’t settled on a common approach to quantum computing, so different research groups have been working on different implementations simultaneously. Numerous small startups have sprung up, each addressing some aspect of a quantum system.
The Quantum Economic Development Consortium (QED-C) is attempting to bring them all together, under the leadership of executive director Celia Merzbacher. So far, QED-C has 132 member companies, which is a lot for an industry still finding its footing. This outpaces the number of universities and independent research centers working on quantum computing, of which there are 36, as well as 11 federally funded research centers.
“In 2018, Congress passed the National Quantum Initiative Act, in which they directed the Department of Commerce, and specifically NIST, to establish a consortium of stakeholders, and that word ‘stakeholders’ was intentionally broad,” said Merzbacher. “The consortium was supposed to look around and figure out what gaps existed — gaps in technology, gaps in research programs that the government might want to fill, gaps in workforce, gaps in standards, anything that might be sort of a gap on the path to a quantum economy of the future.”
The organization provides a means to establish roadmaps and set expectations, and it has many reports available to the public on its website (although specific roadmaps and other documents are reserved for members.)
Why quantum computing is needed
Most people assume it’s the next big thing, and it very well may be. But quantum computing has received so much hype, and so much investment, that some might not even ask this question: ‘What can it actually do?’
As much as quantum computing is discussed, it’s just one of three major targets for quantum technology. The other two are networking and sensing.
Quantum networking leverages superposition, entanglement, and the impossibility of copying a quantum state to create networks that appear unhackable. If two qubits are entangled and one of them moves to another node, then a change to either qubit is immediately reflected in the other one, making such communication practically instantaneous. The big challenge is in maintaining state entanglement over long distances and time.
Quantum sensing addresses specific tasks that it can perform more accurately than non-quantum sensors. These tend to involve gravity and magnetism, as well as inertial changes, benefiting navigation, biomedicine, natural resource mapping, and defense applications.
Quantum computing gets the spotlight
However, it is unlikely that quantum computing will improve everything simply because it can, in some sense, do things in parallel. Speed is only part of the equation. Power matters, too. In isolation, a quantum computer should consume less power than a conventional one. The challenge is cooling.
“Quantum has the potential to offer very low power operation, but now the problem is that you spend lots of energy on the cryogenics because you need a refrigerator to get you down to 1 K,” said Pushkar Apte, strategic technology advisor at SEMI.
Solving the power issues helps, but it isn’t a guarantee, because conventional computing isn’t standing still. Problems that might have benefited from quantum as determined 20 years ago may no longer do so.
“Even though a quantum computer could do the traveling salesman problem perfectly, how much better would that be than these pretty darn good models that we are using today?” Merzbacher asked.
The one certain thing is Shor’s algorithm, which factors large numbers, will undermine the current public-key approach to security. Ridiculously hard problems for regular computers become tractable with quantum computers. It’s for this reason that post-quantum cryptography is often discussed, referring to algorithms that remain secure even in the presence of quantum computers.
Some see quantum computers as accelerators under the direction of conventional computers, ensuring both will have ongoing value.
“I was talking with someone at Oak Ridge [National Lab], probably 10 years ago now, and I asked, ‘What about this quantum thing?’” Merzbacher recalled. “They said, ‘It looks like it will initially be what we think of as an accelerator of high-performance computing.’ We had CPUs, then we added GPUs, and now we’re going to add QPUs.”
Even if a general-purpose quantum computer is developed someday, it might be after specialized systems with more limited functions.
“There’s a lot of talk about fault-tolerant general-purpose quantum computers, but on the way to having those kinds of systems, there will likely be more application-specific platforms with the whole stack designed to do something that the military cares about or drug companies have decided they’re willing to pay for,” Merzbacher said.
Who will have quantum computers?
Conventional computers have infiltrated a vast range of systems we use every day, from tiny units in smartphones, to desktop computers, to the monsters that serve up AI training and inference. Most everyone owns at least one unit, and many own more than one. And even more access those data-center servers from their client units.
Will that eventually be the case with quantum? Highly unlikely, if for no other reason than that the infrastructure necessary for quantum computing (at least today) would preclude its use in small handheld systems unless room-temperature operation is achieved.
Therefore, until and unless we can jettison the big cooling units, quantum computers are likely to remain in data centers of some kind. Only the largest enterprises might consider hosting their own units, with others subscribing to quantum-computing services.
“We’re unlikely to go through a phase where every enterprise has its own quantum computer in a closet somewhere,” said Merzbacher.
How do we measure progress?
As with many engineering problems, there’s no obvious, solid point where we transition from, “Are we there yet?” to “We’re here.” So measuring where we are is necessarily a fuzzy exercise.
“To check progress in quantum hardware, I like a metric that measures how many qubits can be entangled with fidelity greater than 0.5 by applying quantum gates to a simple initial qubit configuration,” said Igor Markov, distinguished architect, engineering architecture at Synopsys. “Entanglement is needed (but not sufficient) for quantum algorithms to do better than conventional algorithms, and quantum fidelity tells us how accurate the result is. 50% is not great accuracy, but being below 50% rules out some uses entirely, whereas fidelity above 50% sometimes can be amplified further.”
Markov pointed to the progress one leading company has made, while still falling short of what’s necessary for commercial deployment. “IBM recently announced entangling 128 superconducting qubits with fidelity above 0.5. This is a good snapshot of where the world stands today. But experiments with neutral atoms [another qubit option] loaded over 10,000 atoms that can be controlled to some extent. This is also very promising.”
Such progress is good, but there is much to do. “The challenge with IBM’s qubits is that because they’re so sensitive to temperature and electrical disturbances, they’re not stable,” Apte said.
Multiple qubit technologies remain in play. “There was a lot of enthusiasm about spin-based systems because they are manufactured using semiconductor fab-type capabilities, and so companies like Intel have pinned a lot of hopes on being able to manufacture a lot of them,” said Merzbacher. “There’s a lot of excitement about atoms and ions.”
Quantum and conventional together
One startup has an arrangement of qubits that resembles a memory, and it’s combining standard and quantum logic on the same chip.
“We’ve got a chip with co-integrated control electronics,” said Maud Vinet, CEO and co-founder at Quobly. “Just like in standard arrays, you’ve got the periphery to go and read the qubits in the array. It operates between 500 mK and 1 K.”
Of course, conventional logic must work at the low temperatures required for quantum. To design that logic, they needed a process design kit (PDK) that accounted for those temperatures.
“We did characterization at low temperature, and that’s the only thing that changed in the PDK,” said Vinet. “Designers can reuse a lot of IP, and they just tweak the dimensions to match the low temperature.”
Still lacking, however, is a solid quantitative way to measure performance. “There’s no analog to the high-performance computing benchmarks such as gigaflops or teraflops or petaflops,” Merzbacher said.
What still needs work to make quantum viable?
It’s well accepted that much more work is necessary to bring quantum technology into full production. The best-known work deals with building qubits. A wide range of approaches is underway, most of which require cooling, although not all require supercooling. There are glimmers of hope for room-temperature qubits, but there’s so much thermal (and other) noise at room temperatures that this will be a big challenge.
“Superconducting qubits operate at 0.04 K, and this is unlikely to change in the foreseeable future,” said Markov. “Some control circuits can run at several K, some at 77 K, and one can go higher at the cost of awkward form factors.”
Temperatures such as 77 K are much easier to achieve with liquid nitrogen than superconducting ones. But they’re still extremely cold.
“This is very, very cold, as compared to room temperature, but so much hotter than superconductors,” remarked Vinet.
“Some other technologies were initially operated at high temperatures, but thermal noise and the need for precise quantum measurements (especially single-photon detectors) are now pushing all leaders into dilution refrigerators,” noted Markov, referring to units that employ differences in helium isotopes to cool their contents down to a few millikelvins.
Achieving a viable, scalable hardware platform will require engineering the qubits, as well as all of the other hardware and infrastructure. First implementations typically serve as proofs of concept. More work is necessary to improve performance and efficiency, and plenty remains.
“There are three areas where advancement is going to have to be made,” said Merzbacher. “One is the hardware, the qubits connecting them, making good qubits. The second thing is error correction, and there have been some important advances in that area. The third leg is algorithms and software.”
It must be reliable
Much of the outstanding work deals with error correction. A distinction now exists between physical and logical qubits. A logical qubit typically comprises several physical qubits and includes some redundant qubits for error correction. It’s analogous to, say, 8B/10B coding where 8 bits of data are represented by a 10-bit code that provides some error protection.
But the specific types of codes to employ haven’t been fully sorted. So-called surface codes predominate now but don’t scale well. Much more work is necessary here for an approach the entire industry can agree on.
The other area requiring enormous investment is software and algorithms. The software isn’t just for the target application. The control loop and all other infrastructural elements also will need software to run. Algorithm work, meanwhile, will help to identify those problems that will best benefit from quantum, along with the best ways to implement the solutions.
Finally, when all research and development is complete, there must be a place to build components at high volume. “I don’t think we are anywhere close to having foundries even a tenth the scale of TSMC,” said Apte. “Quantum production is in R&D-scale facilities that are wholly owned. The quantum industry is in a state of vertical integration, like semiconductors were in the ’80s or ’70s.”
When will we see quantum?
Even given the work ongoing now, some aren’t expecting widespread quantum technology until the 2040s or 2050s, absent some unexpected breakthrough (although, of course, some prognosticators are a decade or so more bullish). That timing partly reflects expectations on when Shor’s algorithm will be practical against 256-bit keys. It’s not that computing itself will take that long to emerge, but rather that we need computing strong enough to employ large numbers of qubits to solve real-world problems. Some say Shor’s algorithm will require millions of qubits to remain robust against noise.
For now, research and development proceed at a stately pace. It may be that some big discovery changes the pace of things, much as LLMs changed the pace of AI development, but, of course, such a thing can’t be predicted (by definition). Improvements to components necessary for a complete system are likely to advance as bits and pieces over time. Innovation will address not just the individual elements but also how they’re combined to create an efficient computing resource.
QED-C queried its membership to assess when they expected a commercial offering (not necessarily a Shor-solving one). “About half said three to five years, and about a third said more than five years,” said Merzbacher. “It did show that there was still a diversity of thinking.”
Some folks see things speeding up. “I was in a conversation with the founder of one of our member companies, and he’s a quantum physicist,” she continued. “He said, ‘If you had asked me two or three years ago how long until we have a useful quantum computer, I would have said 10 years — maybe 15 or more. Now here we are, just a couple years later, and if you ask me this question, I will say three to five years.’ So it does seem that we’re accelerating progress.”
The dust will eventually settle
For now, the primary quantum efforts relate to cryptography, establishing a quantum-safe baseline as soon as possible to forestall the current situation where people download and store encrypted documents in the expectation of being able to decrypt them later. Most other potential uses of the technology will remain in wait-and-see mode for the time being.
“Which ones are interesting science projects, and which of them is likely to actually lead to a company that’s delivering value, we don’t know yet,” Merzbacher added. “There are smart physicists working very diligently at several different ‘transistor’ types — superconducting, photonic, atom-based, ion-based — and it hasn’t become clear which of these modalities is going to get the biggest slice of the pie. And there may be more than one slice.”
But it will be worth waiting for. “Quantum, in my naive view, is standing second in line behind photonics in terms of making a big impact,” said Apte.
Related Articles
Consumer And Med Tech Mushroom As Quantum Closes In
Bio-inspired medical devices can gather more data than ever before, and HPC/quantum will massively speed up discovery rates; consumer self-testing will be slower.
Securing Hardware For The Quantum Era
Quantum computers may become a security threat as early as next year, and that threat will continue to grow over the next several years.
Lines Blurring Between Supercomputing And HPC
Acceleration of performance improvements due to AI and disaggregation are driving significant changes at the leading edge of computing
 |
 |
 |
 |
 |  |
 |  |
 |  |
Leave a Reply