They probably won’t help you crunch numbers on a spreadsheet. They may never run apps as well as the earliest smartphones. And for now, at least, there’s not much they can do better than an inexpensive laptop. But quantum computers are poised to become a significant part of the business technology landscape — and in some industries, they may even turn that landscape upside down. What’s so special about these futuristic machines, and how can businesses prepare for them to reach maturity? Here’s your primer on quantum computing, from its origins in strange natural phenomena to its potential role in data science, machine learning, and more.
From quantum mechanics to quantum computers
The origins of quantum computing date back to the late 1800s and early 1900s, when scientists discovered that atoms and particles behave very differently from human-scale objects. They can tunnel through barriers, for instance. They can exchange information from opposite ends of the universe. And they only possess specific characteristics once they’re measured or observed; until then, they may exist in a combination of all possible states known as a “superposition.” (Some may be familiar with this paradoxical concept from descriptions of the famous Schrödinger's cat thought experiment, which asks us to imagine an unlucky feline sharing a sealed box with a vial of poison gas that may or may not have shattered. According to the bizarre rules of quantum mechanics, the cat must be considered both alive and dead — or neither alive nor dead — until somebody opens the box and determines its fate.)
In the 1980s, several physicists — including Nobel laureate Richard Feynman — suggested that it might be possible to harness the eccentricities of the subatomic world to build more powerful computers. To this day, classical digital computers run on basic units of information called bits that can only have one of two possible states, represented as 0 and 1. But what if bits were designed to have quantum properties, including the ability to be in superpositions of 0 and 1? Then, physicist David Deutsch theorized in 1985, a series of these bits could do many calculations simultaneously, far outperforming the same number of regular bits and maybe even solving problems that classical computers simply can’t in any reasonable amount of time.
It wasn’t until 1998 that the first working quantum computers appeared, featuring just two quantum bits — or qubits, as they’ve come to be called. Since then, researchers have been making important breakthroughs at an increasing pace, funded by major universities, government agencies, and companies like IBM, Google, Microsoft, and Intel. One of their goals is building machines with more qubits than previous models, since a quantum computer’s power grows exponentially with each additional qubit. Another is designing systems that fully isolate qubits from the outside world, allowing them to obey the laws of quantum mechanics. Yet another — and perhaps the most crucial of all — is figuring out what quantum computers do well.
The big three: simulation, optimization, and machine learning
Quantum computers have the potential to work better and faster than classical computers, but that doesn’t mean they always do. And often, it’s impossible to know definitively whether a quantum computer is the right technology for the job. For example, there have been instances when physicists wrote quantum algorithms that appeared to solve problems more efficiently than the best classical algorithms, only to have classical computer scientists later unveil a superior one, says Peter Love, who teaches quantum physics at Tufts University.
Answering the basic question of how these machines could be useful is pretty easy,” he explains. “Finding and producing an example where it is useful is hard.
Answering the basic question of
how these machines could be useful
is pretty easy. Finding and
producing an example where it is
useful is hard.
To justify the astronomical cost of building or owning a quantum computer, companies and organizations want to focus on the small subset of problems that classical computers haven’t been able to solve or that quantum computers might be able to solve exponentially faster. In their quest to pinpoint these problems, researchers have identified three promising use cases for quantum computing:
1. Simulation: "Might be one of the first applications of quantum computing"
When Richard Feynman first described a quantum computer in 1981, he thought it could help humans more accurately represent the natural world. After all, he reasoned, nature’s smallest building blocks behave in ways that only quantum mechanics can explain. Sure enough, over the last couple of years, researchers have simulated molecules and observed their properties by having qubits stand in for electrons, since they naturally entangle just as real electrons would. And in the not-too-distant future, they might be capable of modeling new drugs, chemicals, and materials as well as or better than today’s most powerful supercomputers, according to many researchers.
“You're using one thing to simulate something that looks and acts very much like it,” explains Jarrod McClean, a research scientist at Google. “Because of the naturalness of this connection, a lot of us believe that quantum simulation might be one of the first applications for quantum computing.”
If McClean and others are right, fields like chemistry and materials science will get a boost like never before. Alán Aspuru-Guzik, a professor of chemistry and chemical biology at Harvard University, says that quantum computers will improve his team’s ability to develop new molecules and materials for renewable energy sources.
“In the case of material simulation, we are certain we are going to get an advantage from quantum computers,” he says. “First, we’re going to be able to simulate molecules exactly rather than approximately, which is currently the case with classical computers. And second, quantum computers could help speed up the process so it no longer has to take 10 years to bring a new material to market.”
This enormous potential means that organizations in a number of fields should keep quantum computers on their radar, McClean says. “Companies that are involved with energy and materials, battery problems, drug design, and materials discovery should have a keen interest in this,” he says. “And companies that are at the frontier of experimental energy technology — like fusion — should be very interested.”
2. Optimization: Finding the best solution among alternatives
Imagine that you’re planning a whirlwind tour of 10 cities across Asia. You want to take the shortest possible route from city to city, but you also want to keep travel costs down by getting the best deal on airfare for every leg of the trip. What you have is an optimization problem — a need to find the best solution from among many possible alternatives. It turns out that quantum computers may handle this type of challenge better than their classical counterparts, using algorithms that optimize unstructured problems much more quickly.
Already, the company behind the world’s first commercially available quantum devices — Canada’s D-Wave Systems — has used its machines to help businesses tackle complex optimization problems. “We did a project with Volkswagen this past year where they took data from Beijing taxicabs and developed an algorithm to smooth out traffic in Beijing,” says Vern Brownell, D-Wave’s CEO. “We've also done work with cancer hospitals on optimizing radiology treatment for tumor treatments to minimize the damage to the surrounding tissue and maximize the damage to the tumor itself.”
Optimization problems also crop up in portfolio analysis, risk assessment, bioinformatics, logistics planning, airline scheduling, systems design, and numerous other fields, Brownell says.
D-Wave’s machines rely on a technique called quantum annealing, making them very different from the quantum computers that companies like Google, IBM, Microsoft, and Intel are working to build. Quantum annealers are only designed to solve certain types of problems, most of which fall under the optimization umbrella. And so far, there’s no theoretical proof that they can do so more quickly and accurately than classical computers, according to Love. Nevertheless, Love believes that, over the next 10 to 15 years, the quantum annealing approach stands the greatest chance of delivering solutions that are beyond the capacity of conventional computers.
3. Machine learning and artificial intelligence: More accuracy and faster
In recent years, machine learning and artificial intelligence have emerged from the fringe and made their way to the forefront of enterprise technology. Indeed, 60 percent of business and technology leaders who participated in a 2017 survey said they’ve already implemented a machine learning strategy. From image recognition and natural language processing to recommendation engines and smartbots, machine learning has begun to power core products and services for organizations in nearly every industry.
These technologies may still be in their infancy, but it’s only a matter of time before quantum computers completely disrupt them, experts believe. Researchers have already developed quantum algorithms that could train neural networks used in machine learning with less data and more accuracy, not to mention much faster — quadratically faster, says McClean. Now all they need is a quantum computer that’s powerful and stable enough to run them.
Brownell, whose company has already begun experimenting with machine learning on its quantum computers, thinks this will be the first important use case for quantum computing. “The fusion of quantum computing and AI is going to make a dramatic difference in our ability to build models that make sense of vast amounts of data,” he says. “This will help businesses understand customer patterns and optimize their operations in many different ways.”
Cloudy days ahead for quantum computing
It only works at temperatures just a hair above absolute zero — 200 times colder than outer space. The slightest sound throws off its calculations. Simply turning it on requires a Ph.D. in physics. And it will set you back millions of dollars.
There are many reasons why a quantum computer won’t sit on your desk or power your smartphone anytime in the near future. Instead, the majority of businesses that need quantum computing capabilities will access them on a subscription basis through the cloud, in the same way that companies currently tap into public cloud providers’ massive computational resources to store data and run applications, experts predict.
“Not only do the machines require a fair amount of specialized infrastructure, but most businesses don’t have the necessary in-house expertise to write and adjust quantum algorithms for the problems they want to solve,” McClean says. “Cloud providers will work closely with customers to identify applications for quantum computing and help them develop algorithms, ultimately through user interfaces that do much of the programming work for them.”
In the quantum simulation space, Google and collaborators from several other organizations have already made some headway in closing the gap between domain experts and quantum computer scientists, McClean notes. They recently launched an open-source software package called OpenFermion, which is designed to help chemists and materials scientists adapt conventional algorithms and equations to run on quantum computers. “By giving people tools like these, it really eases the barrier to entry,” McClean says.
Getting ahead of the quantum wave
Given the extraordinary potential for quantum computing to simplify tough challenges and tackle impossible problems, many business and technology leaders want to know how they can begin preparing for the technology’s widespread commercial availability. That’s a difficult question to answer at a time when researchers are still uncovering the best applications for quantum computers and existing machines have yet to demonstrate “quantum supremacy” — the ability to perform calculations that are beyond the capacity of any classical supercomputer.
The most forward-thinking
companies are already well-
positioned to harness quantum
computing because they’re opening
their doors to more computer
scientists and computational
physicists than ever before.
According to Love, the most forward-thinking companies are already well-positioned to harness quantum computing because they’re opening their doors to more computer scientists and computational physicists than ever before. “For my generation of graduate students, your only option in the business world was to work in finance as a quantitative analyst,” he says. “But now, business computing in every industry involves machine learning and optimization and other techniques that have significant overlap with computational physics. So the idea of having a workforce which is more aware of algorithmic issues and how to solve complex problems — that’s something CIOs are probably already thinking about or should be thinking about. And that will be very helpful when we finally get to the point where quantum computers can do something impactful.”
In McClean’s view, getting companies ready for the future of quantum computing will be a two-sided effort. “The people on the quantum side need to produce more introductory material to help get people within traditional models on board,” he explains. “Meanwhile, companies will need to invest in developing quantum computing expertise. Otherwise, you might have a problem cross your desk and never know that if you gave it to a quantum computer, it could really blow everything else out of the water.”