Technology # A Game Plan for Quantum Computing

Some companies may begin to reap gains from quantum computing within five years. Here's what they should know now.

Pharmaceutical companies have an abiding interest in enzymes. These proteins catalyze all kinds of biochemical interactions, often by targeting a single type of molecule with great precision. Harnessing the power of enzymes may help alleviate the major diseases of our time.

Unfortunately, we don’t know the exact molecular structure of most enzymes. In principle, chemists could use computers to model these molecules in order to identify how they work, but enzymes are such complex structures that most are impossible for classical computers to model.

A sufficiently powerful quantum computer, however, could accurately predict in a matter of hours the properties, structure, and reactivity of such substances — an advance that could revolutionize drug development and usher in a new era in health care.

Quantum computers have the potential to resolve problems of this complexity and magnitude across many different industries and applications, including finance, transportation, chemicals, and cybersecurity.

Solving the impossible in a few hours of computing time, finding answers to problems that have bedeviled science and society for years, unlocking unprecedented capabilities for businesses of all kinds — those are the promises of quantum computing, a fundamentally different approach to computation.

None of this will happen overnight. In fact, many companies and businesses won’t be able to reap significant value from quantum computing for a decade or more, although a few will see gains in the next five years.

But the potential is so great, and the technological advances are coming so rapidly, that every business leader should have a basic understanding of how the technology works, the kinds of problems it can help solve, and how she or he should prepare to harness its potential.

Quantum computing is a fundamentally different approach to computation compared with the kinds of calculations that we do on today’s laptops, workstations, and mainframes. It won’t replace these devices, but by leveraging the principles of quantum physics it will solve specific, typically very complex problems of a statistical nature that are difficult for current computers.

Classical computers are programmed with bits (zeros and ones) as data units. Quantum computers use so-called qubits, which can represent a combination of both zero and one at the same time, based on a principle called superposition.

It’s this difference that gives quantum computers the potential to be exponentially faster than today’s mainframes and servers. Quantum computers can do multiple calculations with multiple inputs simultaneously. Today’s computers can handle only one set of inputs and one calculation at a time.

But when you dig into the details of how a quantum computer actually works, you start to understand that many existing challenges must be solved before quantum computers deliver on that potential.

Some of the obstacles are technical. Qubits, for example, are volatile. Every bit in today’s computers must be in a state of one or zero. A great deal of work goes into ensuring that one bit on a computer chip does not interfere with any other bit on that chip.

Qubits, on the other hand, can represent any combination of zero and one. What’s more, they interact with other qubits. In fact, these interactions are what make it possible to conduct multiple calculations at once.

Controlling these interactions, however, is very complicated. The volatility of qubits can cause inputs to be lost or altered, which can throw off the accuracy of results. And creating a computer of meaningful scale would require hundreds of thousands or millions of qubits to be connected coherently. The few quantum computers that exist today can handle nowhere near that number.

Software and hardware companies — ranging from start-ups you’ve never heard of to research institutes to the likes of Google, IBM, and Microsoft — are trying to overcome these obstacles. They’re working on algorithms that bear little resemblance to the ones we use today, hardware that may well wind up looking very different from today’s gray boxes, and software to help translate existing data into a qubit-ready format.

But they have a long way to go. Although quantum computing as a concept has been around since the early 1980s, the first real proof that quantum computers can handle problems too complicated for classical computers occurred only in late 2019, when Google announced that its quantum computer had solved such a calculation in just 200 seconds. But this was more of a mathematical exercise than anything that could be applied to business — the problem had no real-world use at all.

The nature of quantum mechanics also presents obstacles to exponential speed gains. Today’s computers operate in a very straightforward fashion: they manipulate a limited set of data with an algorithm and give you an answer. Quantum computers are more complicated. After multiple units of data are input into qubits, the qubits are manipulated to interact with other qubits, allowing for a number of calculations to be done simultaneously.

That’s why quantum computers are a lot faster than the machines in use today. But those gains are mitigated by the fact that quantum computers don’t deliver one clear answer. Instead, users get a narrowed range of possible answers. In fact, they may find themselves conducting multiple runs of calculations to narrow the range even more, a process that can significantly lessen the speed gains of doing multiple calculations at once.

Getting a range rather than a single answer makes quantum computers sound less precise than today’s computers. That’s true for calculations that are limited in scope, which is one reason quantum computers won’t replace today’s systems. Instead, quantum computers will be used for different kinds of problems — incredibly complex ones for which eliminating an enormous range of possibilities will save an enormous amount of time.

Quantum computers have four fundamental capabilities that differentiate them from today’s classical computers: (1) quantum simulation, in which quantum computers model complex molecules; (2) optimization (that is, solving multivariable problems with unprecedented speed); (3) quantum artificial intelligence, with better algorithms that could transform machine learning across industries as diverse as pharma and automotive; and (4) prime factorization, which could revolutionize encryption.

The best way to understand the business potential of quantum computing is to see how those capabilities could tackle a variety of use cases. Certain industries have specific problems that are particularly well suited to quantum computing.

In total, we’ve reviewed more than 100 nascent use cases and found that they cover a wide range of problems and sectors, including pharmaceuticals, cybersecurity, finance, materials science, and telecommunications. Our research also suggests significant diversity in the development life cycle of these applications, and in the nature of business benefit they could confer.

To paint a richer picture of these dynamics at work, let’s consider four high-potential applications.

Scientists looking to develop new drugs and substances often need to examine the exact structure of a molecule to determine its properties and understand how it might interact with other molecules. Unfortunately, even relatively small molecules are extremely difficult to model accurately using classical computers, since each atom interacts in complex ways with other atoms.

It’s almost impossible for today’s computers to simulate basic molecules that have relatively few atoms — and proteins, to cite just one example, have thousands of them. That’s why today’s scientists are forced to actually create the molecules in question (using synthetic chemistry) to physically measure their properties.

Often the molecule doesn’t work as expected, entailing more synthesis and testing. Each optimization cycle is expensive and time-consuming. This is one reason why developing new drugs and chemicals is such a lengthy process.

Quantum computers are intrinsically well suited to tackle this problem, since the interaction of atoms within a molecule is itself a quantum system. In fact, experts believe that quantum computers will be able to model even the most complex molecules in our bodies. Every bit of progress in this direction will drive faster development of new drugs and other products, and potentially lead to transformative new cures.

Across every industry, many complex business problems involve a host of variables. Where should I place robots on the factory floor? What’s the shortest route for my delivery truck? What’s the most efficient way to deploy cars, motorcycles, and scooters to create a transportation network that meets user demand? How can I optimize the performance and risk of a financial portfolio?

Solving these problems with classical computing is an arduous, hit-and-miss process. To isolate the inputs that drive performance gains or losses, the number of variables that can be shifted in any calculation must be seriously limited. As a result, companies must make one complicated calculation after another, a costly, time-consuming process given the multiplicity of variables.

But, since quantum computers work with multiple variables simultaneously, they can be used first to dramatically narrow the range of possible answers in a very short time. Classical computing can then be called in to zero in on one precise answer, and its work will still seem slow compared with that of quantum. But, since quantum has eliminated so many possibilities, this hybrid approach will drastically cut the time it takes to find the best solution.

It’s possible that quantum computers could speed the arrival of self-driving vehicles. At Ford, GM, Volkswagen, and other car manufacturers, and at a host of start-ups in the new mobility sector, engineers are running hours upon hours of video, image, and LIDAR data through complex neural networks. Their goal: use AI to teach a car to make crucial driving decisions, such as how to take a turn, where to speed up and slow down, and, crucially, how to avoid other vehicles, not to mention pedestrians.

Training an AI algorithm this way requires a set of computationally intensive calculations, which become increasingly difficult as more data and more complex relationships within the variables are added. This training can tax the world’s fastest computers for days or even months.

Since quantum computers can perform multiple complex calculations with multiple variables simultaneously, they could exponentially accelerate the training of such AI systems. It’s not going to happen anytime soon, though. Translating classical data sets to quantum ones is arduous work, and early quantum AI algorithms have resulted in only modest gains.

Quantum computing poses a serious threat to the cybersecurity systems relied on by virtually every company. Most of today’s online-account passwords and secure transactions and communications are protected through encryption algorithms such as RSA or SSL/TLS. These systems make it easy for businesses to create data that can be shared by authorized users while also being protected from outsiders.

Breaking through that encryption requires massive computational power. It’s virtually impossible for today’s computers to solve the math problem behind well-architected encryption quickly enough to be of practical use. (That math problem is known as prime factorization, since encryption is built around the manipulation of large prime numbers.) When data theft does occur, it’s often because of poor implementation of cybersecurity protocols.

Since quantum computers can perform multiple calculations simultaneously, they have the potential to break any classical encryption system. In fact, a quantum algorithm to do just that already exists. (It’s called Shor’s algorithm.) Luckily, there’s no quantum computer capable of managing the hundreds of thousands to millions of qubits it would take to execute Shor’s algorithm — as we said earlier, today’s versions can handle a dozen or so qubits.

But somewhere between 10 and 20 years from now, that might change, and at that point a new wave of quantum encryption technologies would be required to protect even our most basic online services. Scientists — as well as forward-thinking policy makers — are already at work on this quantum cryptography, trying to prepare for this tipping point.

Quantum computing is a complex technology. It’s not an app that’s going to appear one day and be adopted by millions of people the next. After speaking with dozens of experts in the rapidly growing quantum ecosystem, we’ve developed a clear estimate of how the technology will progress over the next couple of decades.

Quantum computers will be expensive machines developed and operated by a few key players. Companies such as Google and IBM hope to double the capabilities of quantum computers, in a Moore’s Law–like fashion, every year. Along with a small but significant cohort of promising start-ups, they will steadily drive up the number of qubits that can be handled by their computers.

Since the technology is nascent, their progress may be slow: our estimate is that by 2030 only 2,000 to 5,000 quantum computers will be operational. Since there are many pieces to the quantum-computing puzzle, the hardware and software needed to handle the most complex problems may not exist until 2035 or beyond.

Nevertheless, quantum will start delivering value to some businesses well before then. Initially, and perhaps in the long term as well, businesses will receive quantum services via the cloud from the same providers they rely on now. Amazon Web Services, Microsoft Azure, and others have already announced quantum offerings. These cloud offerings could quickly expand adoption and demand.

Between 2022 and 2026, we expect many businesses with optimization issues to adopt hybrid approaches, in which parts of the problem would be handled by classical computing and parts by quantum. In that same time frame, quantum computers are likely to become powerful enough to start handling meaningful simulations of molecular structures for chemical, materials, and pharmaceutical companies.

The arrival of quantum AI is further off, and we don’t expect quantum computers to be powerful enough for prime factorization until the very late 2020s at the earliest.

This timeline for the development of the technology informs our estimates of when different industries are likely to benefit most from quantum computing. The experts we spoke with expect that pioneers in advanced industries, global energy and materials, finance, and (to a lesser extent) travel and logistics might start generating significant value from quantum by 2025.

The big payoff for pharmaceuticals may not come until the following decade, given that solving the most complex medical problems involves mimicking deeply complex molecules. By the mid-2030s, a wide range of industries will have the potential to create significant value from quantum computing.

*This article was originally published in *McKinsey Quarterly*. It is republished here by permission.*

*Alex M**énard is a McKinsey partner who leads the firm’s telecommunications, media, and technology practice in France. Ivan Ostojic is a partner and leader of McKinsey’s innovation work in Europe, the Middle East, and Africa. Mark Patel is a senior partner in San Francisco who advises clients on the Internet of Things, analytics, and digital. At the time this article was written, Daniel Volz was a senior management consultant for McKinsey in Germany.*