Quantum Computing: Everything You Need to Know
Accustomed to imagining worst-case scenarios, many cryptography experts are more concerned than usual these days: one of the most widely used schemes for safely transmitting data is poised to become obsolete once quantum computing reaches a sufficiently advanced state.
The cryptosystem known as RSA provides the safety structure for a host of privacy and communication protocols, from email to internet retail transactions. Current standards rely on the fact that no one has the computing power to test every possible way to de-scramble your data once encrypted, but a mature quantum computer could try every option within a matter of hours.
It should be stressed that quantum computers haven’t yet hit that level of maturity — and won’t for some time — but when a large, stable device is built (or if it’s built, as an increasingly diminishing minority argue), its unprecedented ability to factor large numbers would essentially leave the RSA cryptosystem in tatters. Thankfully, the technology is still a ways away — and the experts are on it.
“Don’t panic.” That’s what Mike Brown, CTO and co-founder of quantum-focused cryptography company ISARA Corporation, advises anxious prospective clients. The threat is far from imminent. “What we hear from the academic community and from companies like IBM and Microsoft is that a 2026-to-2030 timeframe is what we typically use from a planning perspective in terms of getting systems ready,” he said.
Cryptographers from ISARA are among several contingents currently taking part in the Post-Quantum Cryptography Standardization project, a contest of quantum-resistant encryption schemes. The aim is to standardize algorithms that can resist attacks levied by large-scale quantum computers. The competition was launched in 2016 by the National Institute of Standards and Technology (NIST), a federal agency that helps establish tech and science guidelines, and is now gearing up for its third round.
Indeed, the level of complexity and stability required of a quantum computer to launch the much-discussed RSA attack is “very extreme,” according to John Donohue, scientific outreach manager at the University of Waterloo’s Institute for Quantum Computing. Even granting that timelines in quantum computing — particularly in terms of scalability — are points of contention, “the community is pretty comfortable saying that’s not something that’s going to happen in the next five to 10 years,” he said.
When Google announced that it had achieved “quantum supremacy” — or that it used a quantum computer to run, in minutes, an operation that would take thousands of years to complete on a classical supercomputer — that machine operated on 54 qubits, the computational bedrocks of quantum computing. While IBM’s Q 53 system operates at a similar level, many current prototypes operate on as few as 20 — or even five — qubits.
But how many qubits would be needed to crack RSA? “Probably on the scale of millions of error-tolerant qubits,” Donohue told Built In.
Scott Aaronson, a computer scientist at the University of Texas at Austin, underscored the same last year in his popular blog after presidential candidate Andrew Yang tweeted that “no code is uncrackable” in the wake of Google’s proof-of-concept milestone.
That’s the good news. The bad news is that, while cryptography experts gain more time to keep our data secure from quantum computers, the technology’s numerous potential upsides — ranging from drug discovery to materials science to financial modeling — is also largely forestalled. And that question of error tolerance continues to stand as quantum computing’s central, Herculean challenge. But before we wrestle with that, let’s get a better elemental sense of the technology.
WHAT IS A QUANTUM COMPUTER?
Quantum computers process information in a fundamentally different way than classical computers. Traditional computers operate on binary bits — information processed in the form of ones or zeroes. But quantum computers transmit information via quantum bits, or qubits, which can exist either as one or zero or both simultaneously. That’s a simplification, and we’ll explore some nuances below, but that capacity — known as superposition — lies at the heart of quantum’s potential for exponentially greater computational power.
Such fundamental complexity both cries out for and resists succinct laymanization. When the New York Times asked 10 experts to explain quantum computing in the length of a tweet, some responses raised more questions than they answered:
Microsoft researcher David Reilly:
“A quantum machine is a kind of analog calculator that computes by encoding information in the ephemeral waves that comprise light and matter at the nanoscale.”
D-Wave Systems executive vice president Alan Baratz:
“If we’re honest, everything we currently know about quantum mechanics can’t fully describe how a quantum computer works.”
Quantum computing also cries out for a digestible metaphor. Quantum physicist Shohini Ghose, of Wilfrid Laurier University, has likened the difference between quantum and classical computing to light bulbs and candles: “The light bulb isn’t just a better candle; it’s something completely different.”
Rebecca Krauthamer, CEO of quantum computing consultancy Quantum Thought, compares quantum computing to a crossroads that allows a traveler to take both paths. “If you’re trying to solve a maze, you’d come to your first gate, and you can go either right or left,” she said. “We have to choose one, but a quantum computer doesn’t have to choose one. It can go right and left at the same time.”
“It can, in a sense, look at these different options simultaneously and then instantly find the most optimal path,” she said. “That's really powerful.”
It’s all about quantum superposition
The most commonly used example of quantum superposition is Schrödinger’s cat:
Despite its ubiquity, many in the QC field aren’t so taken with Schrodinger’s cat. The more interesting fact about superposition — rather than the two-things-at-once point of focus — is the ability to look at quantum states in multiple ways, and ask it different questions, said Donohue. That is, rather than having to perform tasks sequentially, like a traditional computer, quantum computers can run vast numbers of parallel computations.
Part of Donohue’s professional charge is clarifying quantum’s nuances, so it’s worth quoting him here at length:
“In superposition I can have state A and state B. I can ask my quantum state, are you A or B? And it will tell me, ‘I'm a or I'm B.’ But I might have a superposition of A + B — in which case, when I ask it, ‘Are you A or B?’ It’ll tell me A or B randomly.
But the key of superposition is that I can also ask the question, ‘Are you in the superposition state of A + B?’ And then in that case, they'll tell me, ‘Yes, I am the superposition state A + B.
But there’s always going to be an opposite superposition. So if it’s A + B, the opposite superposition is A - B.”
That’s about as simplified as we can get before trotting out equations. But the top-line takeaway is that that superposition is what lets a quantum computer “try all paths at once.”
That’s not to say that such unprecedented computational heft will displace or render moot classical computers. “One thing that we can really agree on in the community is that it won’t solve every type of problem that we run into,” said Krauthamer.
But quantum computing is particularly well suited for certain kinds of challenges. Those include probability problems, optimization (what is, say, the best possible travel route?) and the incredible challenge of molecular simulation for use cases like drug development and materials discovery.
The cocktail of hype and complexity has a way of fuzzing outsiders’ conception of quantum computing — which makes this point worth underlining: quantum computers exist, and they are being used right now.
They are not, however, presently “solving” climate change, turbocharging financial forecasting probabilities or performing other similarly lofty tasks that get bandied about in reference to quantum computing’s potential. QC may have commercial applications related to those challenges, which we’ll explore further below, but that’s well down the road.
Today, we’re still in what’s known as the NISQ era — Noisy, Intermediate-Scale Quantum. In a nutshell, quantum “noise” makes such computers incredibly difficult to stabilize. As such, NISQ computers can’t be trusted to make decisions of major commercial consequence, which means they’re currently used primarily for research and education.
“The technology just isn’t quite there yet to provide a computational advantage over what could be done with other methods of computation at the moment,” said Dohonue. “Most [commercial] interest is from a long-term perspective. [Companies] are getting used to the technology so that when it does catch up — and that timeline is a subject of fierce debate — they’re ready for it.”
Also, it’s fun to sit next to the cool kids. “Let’s be frank. It’s good PR for them, too,” said Donohue.
But NISQ computers’ R&D practicality is demonstrable, if decidedly small-scale. Donohue cites the molecular modeling of lithium hydrogen. That’s a small enough molecule that it can also be simulated using a supercomputer, but the quantum simulation provides an important opportunity to “check our answers” after a classical-computer simulation. NISQs have also delivered some results for problems in high-energy particle physics, Donohue noted.
One breakthrough came in 2017, when researchers at IBM modeled beryllium hydride, the largest molecule simulated on a quantum computer to date. Another key step arrived in 2019, when IonQ researchers used quantum computing to go bigger still, by simulating a water molecule.
These are generally still small problems that can be checked using classical simulation methods. “But it’s building toward things that will be difficult to check without actually building a large particle physics experiment, which can get very expensive,” Donohue said.
And curious minds can get their hands dirty right now. Users can operate small-scale quantum processors via the cloud through IBM’s online Q Experience and its open-source software Quiskit. Late last year, Microsoft and Amazon both announced similar platforms, dubbed Azure Quantum and Braket. “That’s one of the cool things about quantum computing today,” said Krauthamer. “We can all get on and play with it.”
THE BIG POTENTIAL
Quantum computing may still be in its fussy, uncooperative stage, but that hasn’t stopped commercial interests from diving in.
IBM announced at the recent Consumer Electronics Show that its so-called Q Network had expanded to more than 100 companies and organizations. Partners now range from Delta Air Lines to Anthem health to Daimler AG, which owns Mercedes-Benz.
Some of those partnerships hinge on quantum computing’s aforementioned promise in terms of molecular simulation. Daimler, for instance, is hoping the technology will one day yield a way to produce better batteries for electric vehicles.
Elsewhere, partnerships between quantum computing startups and leading companies in the pharmaceutical industry — like those established between 1QBit and Biogen, and ProteinQure and AstraZeneca — point to quantum molecular modeling’s drug-discovery promise, distant though it remains. (Today, drug development is done through expensive, relatively low-yield trial-and-error.)
Researchers would need millions of qubits to compute “the chemical properties of a novel substance,” noted theoretical physicist Sabine Hossenfelder in the Guardian last year. But the conceptual underpinning, at least, is there. “A quantum computer knows quantum mechanics already, so I can essentially program in how another quantum system would work and use that to echo the other one,” explained Donohue.
There’s also hope that large-scale quantum computers will help accelerate AI, and vice versa — although experts disagree on this point. “The reason there’s controversy is, things have to be redesigned in a quantum world,” said Krauthamer, who considers herself an AI-quantum optimist. “We can’t just translate algorithms from regular computers to quantum computers because the rules are completely different, at the most elemental level.”
Some believe quantum computers can help combat climate change by improving carbon capture. Jeremy O’Brien, CEO of Palo Alto-based PsiQuantum, wrote last year that quantum simulation of larger molecules — if achieved — could help build a catalyst “for ‘scrubbing’ carbon dioxide directly from the atmosphere.”
COME CORRECT: CHALLENGES AHEAD
Long-term applications tend to dominate headlines, but they also lead us back to quantum computing’s defining hurdle — and the reason coverage remains littered with terms like “potential” and “promise”: error correction.
Qubits, it turns out, are higher maintenance than even the most meltdown-prone rock star. Any number of simple actions or variables can send error-prone qubits falling into decoherence, or the loss of a quantum state (mainly that all-important superposition). Things that can cause a quantum computer to crash include measuring qubits and running operations — in other words: using it. Even small vibrations and temperature shifts will cause qubits to decohere, too.
That’s why quantum computers are kept isolated, and the ones that run on superconducting circuits — the most prominent method, favored by Google and IBM — have to be kept at near-absolute zero (a cool -460 degrees Fahrenheit).
The challenge is two-fold, according to Jonathan Carter, a scientist at Lawrence Berkeley National Laboratory. First, individual physical qubits need to have better fidelity. That would conceivably happen either through better engineering, discovering optimal circuit layout, and finding the optimal combination of components. Second, we have to arrange them to form logical qubits.
“Estimates range from hundreds to thousands to tens of thousands of physical qubits required to form one fault-tolerant qubit. I think it’s safe to say that none of the technology we have at the moment could scale out to those levels,” Carter said.
From there, researchers would also have to build ever-more complex systems to handle the increase in qubit fidelity and numbers. So how long will it take until hardware-makers actually achieve the necessary error correction to make quantum computers commercially viable?
“Some of these other barriers make it hard to say yes to a five- or 10-year timeline,” Carter said.
Donohue invokes — and rejects — the same figure. “Even the optimist wouldn’t say it’s going to happen in the next five to 10 years,” he said. At the same time, some small optimization problems, specifically in terms of random number generation could happen “very soon.”
“We’ve already seen some useful things in that regard,” he said.
For people like Michael Biercuk, founder of quantum-engineering software company Q-CTRL, “the only technical commercial milestone that matters now is quantum advantage” — or, as he uses the term, when a quantum computer provides some time or cost advantage over a classical computer. Count him among the optimists: he foresees a five-to-eight year time scale to achieve such a goal.
Another open question: Which method of quantum computing will become standard? While superconducting has borne the most fruit so far, researchers are exploring alternative methods that involve trapped ions, quantum annealing or so-called topological qubits. In Donohue’s view, it’s not necessarily a question of which technology is better so much as one of finding the best approach for different applications. For instance, superconducting chips naturally dovetail with the magnetic field technology that underpins neuroimaging.
The challenges that quantum computing faces, however, aren’t strictly hardware-related. The “magic” of quantum computing resides in algorithmic advances, “not speed,” Greg Kuperberg, a mathematician at the University of California at Davis, is quick to underscore.
“If you come up with a new algorithm, for a question that it fits, things can be exponentially faster,” he said, using exponential literally, not metaphorically. (There are currently 63 algorithms listed and 420 papers cited at Quantum Algorithm Zoo, an online catalog of quantum algorithms compiled by Microsoft quantum researcher Scott Jordan.)
Another roadblock, according to Krauthamer, is general lack of expertise. “There’s just not enough people working at the software level or at the algorithmic level in the field,” she said. Tech entrepreneur Jack Hidarity’s team set out to count the number of people working in quantum computing and found only about 800 to 850 people, according to Krauthamer. “That’s a bigger problem to focus on, even more than the hardware,” she said. “Because the people will bring that innovation.”
While the community underscores the importance of outreach, the term “quantum supremacy” has itself come under fire. “In our view, ‘supremacy’ has overtones of violence, neocolonialism and racism through its association with ‘white supremacy,’” 13 researchers wrote in Nature late last year. The letter has kickstarted an ongoing conversation among researchers and academics.
But the field’s attempt to attract and expand also comes at a time of uncertainty in terms of broader information-sharing.
COLLABORATION VS. CONTROL
Quantum computing research is sometimes framed in the same adversarial terms as conversations about trade and other emerging tech — that is, U.S. versus China. An oft-cited statistic from patent analytics consultancy Patinformatics states that, in 2018, China filed 492 patents related to quantum technology, compared to just 248 in the United States. That same year, the think tank Center for a New American Security published a paper that warned, “China is positioning itself as a powerhouse in quantum science.” By the end of 2018, the U.S. passed and signed into law the National Quantum Initiative Act. Many in the field believe legislators were compelled due to China’s perceived growing advantage.
The initiative has spurred domestic research — the Department of Energy recently announced up to $625 million in funding to establish up to five quantum information research centers — but the geopolitical tensions give some in the quantum computing community pause, namely for fear of collaboration-chilling regulation. “As quantum technology has become prominent in the media, among other places, there has been a desire suddenly among governments to clamp down,” said Biercuk, who has warned of “poorly crafted” and “nationalistic” export controls in the past.
“What they don’t understand often is that quantum technology — and quantum information in particular — really are deep research activities where open transfer of scientific knowledge is essential,” he added.
The National Science Foundation — one of the government departments given additional funding and directives under the act — generally has a positive track record in terms of avoiding “draconian” security controls, Kuperberg said. Even still, the antagonistic framing tends to obscure the on-the-ground facts. “The truth behind the scenes is that, yes, China would like to be doing good research and quantum computing, but a lot of what they’re doing is just scrambling for any kind of output,” he said.
Indeed, the majority of the aforementioned Chinese patents are quantum tech, but not quantum computing tech — which is where the real promise lies.
The Department of Energy has an internal list of sensitive technologies that it could potentially restrict DOE researchers from sharing with counterparts in China, Russia, Iran and North Korea. It has not yet implemented that curtailment, however, DOE Office of Science director Chris Fall told the House committee on science, space and technology and clarified to Science, in January.
MONEY, HYPE & MOMENTUM
Along with such multi-agency-focused government spending, there’s been a tsunami of venture capital directed toward commercial quantum-computing interests in recent years. A Nature analysis found that, in 2017 and 2018, private funding in the industry hit at least $450 million.
Still, funding concerns linger in some corners. Even as Google’s quantum supremacy proof of concept has helped heighten excitement among enterprise investors, Biercuk has also flagged “the beginnings of a contraction in investment in the sector.”
Even as “exceptional cases” dominate headlines — he points to PsiQuantum’s recent $230 million venture windfall — there are lesser-reported signs of struggle. “I know of probably four or five smaller shops that started and closed within about 24 months; others were absorbed by larger organizations because they struggled to raise,” he said.
At the same time, signs of at least moderate investor agitation and internal turmoil have emerged. The Wall Street Journal reported in January that much-buzzed quantum computing startup Rigetti Computing saw its CTO and COO, among other staff, depart amid concerns that the company’s tech “wouldn’t be commercially viable in a reasonable time frame.”
Investor expectations had become inflated in some instances, according to experts. “Some very good teams have faced more investor skepticism than I think has been justified … This is not six months to mobile application development,” Biercuk said.
In Kuperberg’s view, part of the problem is that venture capital and quantum computing operate on completely different timelines. “Putting venture capital into this in the hope that some profitable thing would arise quickly, that doesn’t seem very natural to me in the first place,” he said, adding the caveat that he considers the majority of QC money “prestige investment” rather than strictly ROI-focused.
But some startups themselves may have had some hand in driving financiers’ over-optimism. “I won’t name names, but there definitely were some people giving investors outsize expectations, especially when people started coming up with some pieces of hardware, saying that advantages were right around the corner,” said Donohe. “That very much rubbed the academic community the wrong way.”
Scott Aaronson recently called out two prominent startups for what he described as a sort of calculated equivocation. He wrote of a pattern in which a party will speak of a quantum algorithm’s promise, “without asking … whether there are any indications that your approach will ever be able to exploit interference of amplitudes to outperform the best classical algorithm.”
And, mea culpa, some blame for the hype surely lies with tech media. “Trying to crack an area for a lay audience means you inevitably sacrifice some scientific precision,” said Biercuk. (Thanks for understanding.)
It’s all led to a willingness to serve up a glass of cold water now and again. As Juani Bermejo-Vega, a physicist and researcher at University of Granada in Spain, recently told Wired, the machine on which Google ran its milestone proof of concept is “mostly still a useless quantum computer for practical purposes.”
Bermejo-Vega’s quote came in a story about the emergence of a Twitter account called Quantum Bullshit Detector, which decrees, @artdecider-like, a “bullshit” or “not bullshit” quote tweet of various quantum claims. The fact that leading quantum researchers are among the account’s 9,000-plus base of followers would seem to indicate that some weariness exists among the ranks.
But even with the various challenges, cautious optimism seems to characterize much of the industry. “For good and ill, I’m vocal about maintaining scientific and technical integrity while also being a true optimist about the field and sharing the excitement that I have — and to excite others about what’s coming,” Biercuk said.
This year could prove to be formative in the quest to use quantum computers to solve real-world problems, said Krauthamer. “Whenever I talk to people about quantum computing, without fail, they come away really excited. Even the biggest skeptics who say, ‘Oh no, they’re not real. It’s not going to happen for a long time.’”