Quantum Computing for Business Leaders: Will the reality live up to the hype?

In 1994, mathematician Peter Shor introduced a quantum-computing algorithm that could reduce the time it takes to find the prime factors of large numbers from billions of years using a conventional transistor-based computer to a few days using a quantum computer. This was an enormous breakthrough, because prime factorization is the foundation for much of our present encryption and information security infrastructure. Seven years later, IBM scientists successfully demonstrated the algorithm on a quantum machine—albeit a very small one—for the first time, proving that quantum computers could be built and that Shor’s algorithm could be implemented.

Quantum computers solve many problems exponentially faster and with less energy consumption than classical, or binary, computers. To understand why, imagine a two-dimensional maze. A classical computer needs to run one path after the other until it finds the way out of the maze. If the maze comprises 256 possible paths, the classical computer has to run through the maze about 128 consecutive times (on average, half of a maze’s paths must be tried to find the right one). A quantum computer, however, is able to work with all 256 paths at once. To put it a bit differently, an 8-bit classical computer can represent only a single number from 0 to 255, but an 8-qubit quantum computer can represent every number from 0 to 255 simultaneously. How is that possible? The answer is based in fundamental laws of quantum mechanics: While a classical-computing binary unit, or bit, can hold a value of either 0 or 1, a qubit (short for quantum bit) can represent 0 or 1—or it can hold both values at the same time.

Quantum computing will enable businesses to better optimize investment strategies, improve encryption, discover products, and much more. Tremendous levels of investment, private-sector competition, and mathematical and scientific talent are currently being devoted to quantum research. Venture capital funding in the space grew by 500% from 2015 to 2020, according to CB Insights. PsiQuantum, a quantum-computing start-up founded in 2016, has already raised more than $665 million, including investments from BlackRock and Microsoft. Research and development heavyweights Honeywell, IBM, and Intel are also in the race to deliver the next quantum breakthrough. Consulting firms are building deep talent pools to support clients; Accenture has more than 15 teams and 100 experts focused on quantum globally. (Disclosure: Accenture provides financial support to MIT’s Initiative on the Digital Economy, where two coauthors of this piece work.) In May 2021, Google committed to spending several billion dollars to build a functional quantum computer by 2029, and its new campus for quantum AI in Santa Barbara will house hundreds of quantum-dedicated employees, a quantum data center, research labs, and quantum-processor-chip-fabrication facilities.

This is the type of environment that so often in the past has produced breakthrough advances in technology. And make no mistake: The quantum-computing breakthrough will be a big one. It will bring two huge, simultaneous, and sudden changes to the modern business world: The first is an end to our current infrastructure for ensuring digital privacy and security over public networks, leaving companies that have not upgraded their infrastructure wide-open to devastating attacks. The second change is much more positive; it’s an explosion of algorithmic power that will let us do things with computers that are impossible today and that hold the promise to reshape our world.

When will a commercially relevant quantum computer be available? Almost 20 years have passed since the proof-of-principle demonstration of Shor’s algorithm, and scientists continue to face myriad challenges in developing large-scale quantum computers. Skeptics argue that it is too early to get excited—or anxious, depending on your point of view—about quantum computing’s real-world applications. It’s instructive to recall that the transistor was invented in 1947, yet the first 4-bit processor was not introduced for another 25 years, and it was another 25 years after that before Intel introduced the Pentium Pro chip with millions of transistors. Hard tech takes time, and quantum is no exception.

But quantum is coming, and it’s not too soon for business managers to be thinking about how it will spur digital investment, reshape industries, and spark innovation. It won’t make or break your business in the near term, but a solid understanding of quantum applications today is crucial for positioning your company to reap the benefits—and avoid potential catastrophe—during the next decade.

What Is a Quantum Computer?

The principles of quantum mechanics—the science of how matter and light behave at an atomic and subatomic level—are at the core of innovations such as MRI imaging, lasers, atomic clocks, and nanoscale microscopes. But utilizing those principles to build computers requires that we master an entirely new skill: precisely controlling the behavior of quantum systems while preserving their “weird” quantum-mechanical nature. This is a daunting task because quantum systems—like photons and electrons—are very delicate and unstable, and their behavior defies our ingrained view of how the physical world operates. But when harnessed correctly, their counterintuitive forces are features, not bugs, in unlocking new capabilities.

One of the most formidable obstacles to building functional quantum computers is that qubits don’t stick around very long. Vibration, temperature, and other environmental factors can cause them to lose their quantum-mechanical properties, resulting in errors. Today, the rate at which errors occur in qubits limits the duration of algorithms that can be run. Scientists are working to build environments in which many physical qubits act together to create error-protected logical qubits, which can survive for much longer periods of time—long enough to support commercially viable applications. It will most likely take some 1,000 physical qubits to make a single logical qubit; the most advanced quantum computers today have only 50 to 100 physical qubits.

In the past couple of years corporations have become much more involved in building quantum computers. Both IBM and Google, two of the most optimistic technology companies in this space, believe a logical qubit will be demonstrated within two years. As with transistor-based computing, commercial viability will not occur all at once but rather will grow steadily as the number of logical qubits increases and error rates come down.

How Businesses Can Use Quantum Computers

Few firms will build or own quantum computers in the near term. Instead, we’ll see a cloud-computing-style model where companies rent access to quantum machines hosted by a relatively small number of specialist providers, similar to how companies today purchase computing from AWS, Google Cloud, and Microsoft Azure. (Disclosure: Research by coauthor William Oliver has been supported by these and other companies mentioned in this piece.) Quantum computers will not be used in isolation but will be part of a hybrid solution in which tasks will be assigned to the most suitable machine (quantum or classical). A quantum-computing cloud infrastructure will enable the sharing of resources and create economies of scale that lower costs and increase access, which in turn will drive demand and accelerate progress.

As quantum hardware and software improve, algorithm designers will be empowered to experiment and iterate on their ideas and hunches. They’ll be able to refine existing algorithms and create new ones without having to wait years between development and testing on a functional machine.

We may be able to better fight global warming if quantum simulations can tackle materials-science problems, such as finding compounds for more-efficient batteries.

Quantum algorithms are very different from the algorithms classical computers use. Those most likely to apply to business processes fall into five families; some of them will allow us to do standard tasks much more quickly, while others will let us seize entirely new opportunities.

1. Simulation. When quantum pioneers such as Richard Feynman and Paul Benioff first envisioned the quantum computer, they believed that it would unlock secrets to how nature works. We are starting to bear witness to their vision. For example: The modeling of a chemical reaction with 100 strongly correlated electrons (nitrogen fixation is one such reaction) is out of the reach of powerful classical computers. But in 2017, a team led by Markus Reiher, a professor of theoretical chemistry at ETH Zurich, calculated the scale of the quantum system needed for the task and introduced a viable approach. The team found that the goal is achievable using a cluster of advanced machines of about 100 logical qubits each. Examples of the breakthroughs that could emerge from the modeling of natural processes abound. Here are three:

Chemistry. In the early 1900s, Fritz Haber and Carl Bosch developed an industrial process for nitrogen fixation that synthesizes ammonia directly from nitrogen and oxygen—a process still used today to produce fertilizer for crops that feed billions of people around the world. Incredible as the discovery was more than a century ago, it has come at a steep cost: The Haber-Bosch process is now responsible for 1% to 2% of global energy consumption and 1.4% of CO2 emissions. We can do better, and quantum computing can help.

For example, we know that a naturally occurring enzyme can achieve the same results as the Haber-Bosch process while expending only a fraction of the energy. Unfortunately, the limitations of classical computers prevent us from modeling the exact chemical reactions that the enzyme uses. A quantum computer will one day be able to do so, thereby creating new opportunities for chemical companies to produce fertilizer and other products in more energy-efficient ways.

Energy. A type of nuclear fusion known as inertial confinement fusion uses powerful lasers to compress tiny pellets of fuel, generating extremely high temperatures under the right conditions. In theory, the amount of energy released from this process could be greater than that used by the lasers, making it a viable energy source. Achieving this in practice, however, depends on configuring the vast number of possible parameters of the process with incredible precision—something classical computers have done with only limited success. Google engineering director Hartmut Neven believes quantum computing can aid in the design of better reactors, opening up the potential for an abundant form of clean energy.

Life sciences. In 2018 three Harvard chemists published a paper outlining the potential of quantum computing in drug discovery. They detailed how the technology could yield substantial progress by enabling faster and more-accurate characterization of molecular systems. The same year, the researchers cofounded Zapata, a quantum-computing start-up that has since raised more than $65 million in venture capital.

It’s not just start-ups that are looking for new molecules using computers rather than test tubes. QuPharm is a consortium of 17 pharma companies, including AbbVie, Bayer, GSK, Takeda, and Pfizer, that are pooling expertise to accelerate progress in quantum hardware and software. In 2019, biotech firm Biogen and Canadian quantum-computing specialist 1QBit collaborated to develop a quantum-enabled molecular comparison tool, an important part of virtual-screening experiments deployed during the early stages of drug discovery.

Other researchers are examining how quantum could provide new insights into chemical mechanisms such as photosynthesis. We may also be able to better fight global warming if quantum simulations can tackle materials-science problems, such as finding compounds for more-efficient batteries, better solar cells, and new kinds of power lines that transmit energy more efficiently.

2. Linear systems. Equations of linear systems are at the core of many classical-computing applications in engineering, finance, chemistry, economics, and computer science. Quantum computing offers the potential for an exponential improvement in sampling solutions to such equations. (We already know of one such algorithm, called HHL, codeveloped by some of our MIT colleagues.) The most promising linear-systems applications may be in the area of enhanced machine learning. There has been an explosion in the adoption of neural networks—a means of training a computer to perform a task inspired by the way the human brain works—to power a wide variety of applications. This has been accompanied by an increasing need for enhanced training of the computer models.

Take recommendation systems, for example. Netflix models preferences in a large matrix across all its subscribers for all the movies in its archive. Its goal is to recommend films to users that they have not watched before. A quantum algorithm may be able to make such recommendations faster and more accurately than classical computers can, particularly when the number of dimensions in the matrix is very large.

Another linear-systems application could be improving the ability of AI to derive useful information from photos and videos. Researchers from leading quantum firms, for example, recently published a paper detailing how quantum computers might work with classical computers to create original images and videos. In a demonstration, the system produced high-resolution images of handwritten numbers using a machine-learning technique called generative adversarial networks (GANs). Although the output may sound rudimentary today, imagine a future Pixar movie in which elements of a fictional world are created and organized not by graphic designers but by quantum computers. Applications of quantum GANs could include generating 3D objects in architecture and building synthetic DNA data in genomics research to produce novel molecules for cancer treatments.

One of the challenges facing linear-systems algorithms—and other types, as we shall see—is what’s known as the data-loading problem: how to transfer large amounts of classical data into quantum computers. Solving that problem will mark a significant milestone in commercial viability.

3. Optimization. Algorithms for optimization determine which decision in a given scenario is most likely to attain a specific objective. An investment manager, for example, tries to find the optimal retirement strategy for a client by balancing expected returns with some measure of risk. Quantum optimization algorithms can improve the quality of the solutions and increase the computational speed in finding them.

In May 2021, Zapata announced the results of research it conducted with Spanish bank BBVA to investigate the practical application of a quantum system in creating credit valuation adjustments (CVAs)—a regulatory requirement put in place to minimize systemic financial risk. The project focused on a Monte Carlo simulation, the standard technique for CVA risk analysis. The calculations underlying the simulations are complex and time-consuming for classical computers because they must account for a wide range of possible credit-default scenarios. Zapata and BBVA’s research identified the potential for speedups over classical machines as the error-correction rates in future generations of quantum computers improve. Large banks are already investing in the space: Goldman Sachs and JPMorgan Chase, as well as BBVA, have entire teams dedicated to researching the possibilities of quantum computing in banking and finance.

Optimization algorithms benefit companies in a wide range of other industries. Any business that depends on finding the best supply-chain routes or increasing the productivity of a manufacturing facility already knows the importance of optimization in improving performance. Most optimization problems can be adequately solved using classical computers and algorithms. Imagine that you want to optimize your 20-mile drive home from work. Google Maps can approximate the best route without exhaustively trying every single alternative. Whether it chooses the absolute best route or one within a minute of it will not make much difference to you. But for much-larger-scale challenges and those for which incremental improvements are immensely valuable, quantum-computing optimization algorithms may be a game changer.

4. Unstructured search. When a classical computer needs to find an exact information target in an unstructured database, it must search line by line until it finds a query match. But each search result the computer generates gives it no additional information; that is, negative results do not narrow down the possibilities for subsequent searches. This is one of the most basic computer science problems. To find the information faster, one can run multiple classical computers, each searching line by line. With quantum computing, searches can be conducted faster and across larger swaths of data. Applications that rely on database probing include internet search engines, real-time processing of credit-card transactions, and even scanning of astronomical radio waves for signs of extraterrestrial intelligence.

Grover’s algorithm is a powerful quantum search theory developed in 1996 that could dramatically improve the way computers find information in a large unstructured database, solving what’s known as the “needle in a haystack” challenge. Consider genomic technologies, which have provided transformative insights into microbiology—for example, identifying genetic cardiac disorders and offering great potential for real-time detection and surveillance of epidemics. These technologies need lots of computer power. Every time researchers map a DNA sequence to a reference genome, they must perform a massive search on classical computers. Grover’s algorithm could greatly accelerate the speed of these searches, but they can be run only on a functional quantum computer.

In addition to these challenges, unstructured-data algorithms come up against the data-loading problem because they rely on the efficient input of large amounts of classical data into quantum computers.

5. Factoring and encryption. As we’ve discussed, prime factorization underpins much of current global internet security and privacy infrastructure. Bank balances, Bitcoin, credit cards, social media passwords, and just about everything else of interest to cybercriminals is protected by factoring problems that classical computers can’t solve with brute force.

Quantum computing could upend this paradigm, making it much easier to crack the encryption systems we rely on today. In April 2021, the National Institute of Standards and Technology (NIST), the U.S. government body tasked with developing cybersecurity standards, warned that “we cannot predict when a quantum computer capable of executing Shor’s algorithm will be available to adversaries, but…when that day comes, all secret and private keys that are protected using the current public-key algorithms—and all available information protected under those keys—will be subject to exposure.”

Nefarious actors may not be able to crack current encryption, but they can easily acquire data in an encrypted format (by, for example, hacking into an internet service provider and copying all the traffic that flows through it). Imagine if a hacker were to acquire and store encrypted data until a sufficiently sophisticated quantum computer was able to break the encryption. At that point, all the data would be exposed. To prevent such scenarios from happening, the transition to quantum-resistant cryptography must occur long before large-scale quantum computers are operational.

Cybersecurity techniques called postquantum cryptography that can be deployed by classical computers are currently being developed. NIST launched a public competition in 2016 to source algorithms that could potentially resist a quantum computer’s attack. It is due to announce the results in 2022, but even when postquantum algorithms are identified, the process for deploying the new cryptosystems will require massive upgrades to software, hardware, and communications infrastructure. All existing sensitive data will have to be re-encrypted, and new infrastructure will need to be built to support new cryptographic algorithms.

Such remediation efforts will have a significant economic impact. Carl Dukatz, the quantum lead at Accenture, believes that the disruption caused by the move to postquantum cryptography will eclipse the work done to mitigate the Y2K problem, a process that cost the United States and its businesses more than $100 billion. The transition away from quantum-vulnerable infrastructure needs to start years in advance of the arrival of large-scale quantum computers. It’s easy to imagine that before too long companies will have to demonstrate to regulators or auditors that they’re on track to being “quantum compliant,” just as they had to show Y2K compliance in the late 1990s.

Luckily, the arrival of quantum computing isn’t all risk, expense, and downside. It will bring about advances we can’t yet foresee and offer such rich opportunities that the security and encryption transition accompanying the dawn of the quantum era will be one of its least-significant chapters.

What Managers Should Do Now

Even though a commercially viable quantum computer is not yet available, it’s not too soon to get prepared. Managers should focus on two key activities: vigilance and visioning.

Vigilance means keeping an eye on how quickly progress is being made toward key technological milestones. These include the demonstration of the first logical qubit, reductions in error rates, and a proven commercial—not just technical—quantum advantage over classical computers. Companies can track progress using sources such as expert panels and forecasting tournaments. In the months and years ahead, we may find that forecasts have been too conservative and that the quantum era will be here sooner than we thought. But if milestones remain stubbornly hard to reach, then classical dominance will continue for some time.

Visioning, or coming up with plans and scenarios for how quantum computing will affect your company, goes hand in hand with vigilance. In the short term, you should have in place a team of people who understand the implications of quantum computing and can identify the company’s future needs, opportunities, and potential vulnerabilities.

As managers begin to wrap their brains around quantum computing and how it will affect their organizations, they should ask themselves the following questions: Where are we currently constrained by limits to computing capability, and are those areas amenable to any of the five families of quantum algorithms? What are our main uses for machine learning and other types of AI, and how much will quantum computing help in those areas? Finally, what biological or chemical processes would we like to be able to model at the foundational level?

. . .

Unlocking nature’s secrets was the first use of quantum computing envisioned by the field’s pioneers, and it remains the most exciting one. Sometime in the first half of this century, we’ll put qubits to work on that challenge—and many, many others.

Copyright 2022 Harvard Business School Publishing Corp. Distributed by The New York Times Syndicate.