The race is on to develop commercial quantum computers. The breakthroughs they promise — new ways of simulating materials, optimizing processes, and improving machine learning — could transform society, just as today’s digital computers have done. But the route to delivering economic benefits is uncertain. The digital revolution took decades and required businesses to replace expensive equipment and completely rethink how they operate. The quantum computing revolution could be much more painful.
Quantum computers operate in a completely different way from digital computers, and can potentially store and analyze information more efficiently. Digital computers essentially use on–off switches and process binary ‘bits’ of information (0s and 1s). Quantum computers encode information in the quantum state of atoms, electrons, and photons, known as qubits. These qubits can represent many states at once and be combined or ‘entangled,’ thereby speeding up calculations.
In the long run, businesses adopting quantum computing should have a competitive edge over others. Yet, in the short term, it’s unclear to what extent the introduction of these machines will prove commercially valuable.
When digital computers started to gain popularity in the 1970s and 1980s, rather than delivering efficiencies, for 15 years, they slowed growth in productivity, the value added relative to inputs such as labor, by 0.76 percentage points per annum. Such a dip is known as the productivity paradox. It arose because businesses had to invest in new equipment and learn how to program the devices, as well as work out what to use them for. At first, firms did not invest enough in other innovations that were needed to change core processes and business models2. Only after many sectors had adjusted in the 1990s did productivity growth rise again, sharply (see ‘Productivity paradox’).
For example, it took a decade of investment, throughout the 1980s, for large firms, such as the retail corporation Walmart, to routinely process data to coordinate planning and to forecast and replenish their inventory along their supply chains. Walmart gave suppliers access to its sales and inventory data, helping to reduce costs from underproduction or overproduction. The company became able to handle its own distribution and achieve efficiency through economies of scale. All these changes took time and required coordination across many firms.
We think that the quantum computing revolution could lead to an even more severe and expensive learning curve, for three reasons: high integration costs and few short-term rewards; difficulty in translating quantum concepts for business managers and engineers; and the threat to cryptography posed by quantum computers. As a consequence, assuming that the productivity growth rate slows by 50% more than it did for simpler digital computers, we estimate that the introduction of commercial quantum computers could result in economic losses in the gross domestic product (GDP) per capita of approximately US$13,000 over 15 years (based on 2022 levels), or $310 billion per annum in the United States alone.
Fortunately, there are ways to lighten the load and accelerate the benefits to society, three of which we outline here.
Firms might initially adopt quantum computers to solve existing business problems, for which improvements are likely to be incremental. But for more-ambitious uses, the extra costs and likelihood of potential failures might make firms risk-averse. For example, a company that collects vast amounts of data from sensors to inform disaster relief and recovery might look to quantum computers to process information more quickly, to help save lives. But the first such computers might be more prone to faults and errors than digital ones, with potentially grave consequences for life-critical operations. Such companies might therefore be put off from using quantum computers until they are more reliable.
These computers will also need to be networked with digital computers, and integrating two different technologies will be difficult and expensive. Firms will still need digital computers to perform everyday tasks and computations; they will use quantum computers to solve more complex and specialist problems. Yet, developing hybrid protocols and programs that can work in both situations is much harder than it was to program digital computers in the 1970s.
Hybrid systems will need to be fluent in both digital bits and quantum qubits, and able to encode classical data into quantum states and vice versa. They will need converters to translate digital and analog signals to transfer information between the two types of processing units. Quantum computers are generally large and might need to be cryogenically cooled, making it unlikely that many companies will have a machine of their own. Many will buy services remotely in the cloud through the Internet, for example, sourcing extra computing power for simulating materials. Some users, such as traders in financial markets, in which millisecond timing is crucial, might need to host both types of computers.
To bring firms on board quickly, the commercial advantages will need to be demonstrated in practice. For this, government funding will be needed to attract private investment. We suggest this could be framed as a mission to help companies apply quantum computing to industrial and societal grand challenges. For example, for weather forecasting, quantum systems could analyze huge amounts of data to keep up with rapidly changing conditions. The resilience of the financial system could be improved through better modeling of markets, as would the development of low-carbon technologies to address climate change, such as catalysts for carbon capture or electrolytes for batteries.
Economists will need to devise a framework for evaluating the financial benefits of quantum computing, to encourage firms to invest. Researchers should build proof-of-concept cases, starting by identifying areas in which quantum computers might outperform digital computers for societal grand challenges. Researchers should also set out what firms need to do to adopt quantum technologies, including how they might need to change their business models and practices, as well as working with others along their value chains.
Quantum technologies operate on principles that are often counterintuitive and outside the comfort zone of many engineers and business managers. For example, these technologies work probabilistically and don’t seem to obey classical conceptions of cause and effect. According to some schools of thought, in the quantum world, the human agency might influence outcomes, meaning the person operating the computer might need to be considered as part of the system.
And, at present, there’s no shared language among scientists, engineers, and business managers around quantum computing. Misunderstandings and confusion create delays and, therefore, further costs. Managers and engineers will need to know enough to be able to select the right class of problems for quantum computers, know what type of information is required to solve them and prepare data in a quantum-ready format
For example, a delivery logistics company might wish to reschedule its vehicle routes more rapidly to respond better to customer demand for pickups of goods that need returning. Quantum computation could be effective for such replanning — which involves solving a complex combinatorial problem — in which one change has a knock-on effect on other areas of the business, such as inventory management and financing. But managers would need to be able to spot areas of advantage such as this and know what to do to implement quantum computing solutions.
A common semantic and syntactic language for quantum computers needs to be developed. It should be similar to the standardized Unified Modeling Language used for digital computer programming — a visual language that helps software developers and engineers to build models to track the steps and actions involved in business processes. Such a tool reduces the costs of software development by making the process intuitive for business managers. Quantum computers also require algorithms and data structures, yet quantum information is much richer than classical information and more challenging to store, transmit, and receive6.
A quantum unified modeling language that is similar to the classical one but can also work with quantum information will enable scientists, engineers, and managers to stay on the same page while they discuss prototypes, testbeds, road maps, simulation models, and hybrid information-technology architectures7. Design toolkits that consist of reusable templates and guidelines containing standard modules for hardware and software development, will allow users to innovate for themselves, shortening development times.
Some of this is beginning to happen. For example, modular workflows are emerging that enable computational chemists and algorithm developers to customize and control chemistry experiments using early versions of quantum computing platforms. A more concerted approach to standardize the language across application areas and hardware platforms is needed to foster commercialization.
Strategies for communicating about quantum computing with the public are also needed, to build trust in these new technologies and ensure that benefits accrue to all parts of society in a responsible manner. Scientists, policymakers, and communications specialists should work together to create narratives around the usefulness of quantum technologies. They should focus on practical problems that can be solved rather than tales of weird quantum behavior.
Although some such initiatives are being set up as part of national quantum programs, more research is needed to better understand how cognitive biases and ways of learning might influence the adoption of quantum computing. For example, how were cognitive barriers overcome in adopting digital computers and nanotechnologies? Answers to questions such as this will help researchers to develop communication protocols and toolkits.
Quantum computing threatens to break a widely used protocol for encrypting information. Today, sensitive data are typically encrypted by using digital keys in the form of factors for large prime numbers, and sent through fiber-optic cables and other channels as classical bits — streams of electrical and optical pulses representing 1s and 0s. The encryption relies on the inability of classical computers to compute the factors for the prime numbers in a reasonable time. However, quantum computers could, in principle, work out these factors faster and therefore break the encryption.
Addressing this risk will bring further costs. To protect the security of data and communications, firms will need to invest in new mathematical approaches for encryption or use quantum-based communications systems, such as quantum key distribution. Quantum key distribution uses qubits sent either through fiber-optic cables or free space (through the air, vacuum, or outer space), to randomize the generation of keys between the sender and receiver using the probabilistic principles of quantum mechanics. Because of the fragile nature of qubits, if a hacker tries to observe them in transit, the quantum state is affected, and the sender and receiver will know that it was tampered with.
Such a threat to sensitive government data and communications8 could also raise geopolitical issues and lead to export controls, such as those imposed by the United States and the Netherlands on microprocessors. The technology bottlenecks for quantum computing are unclear because there are several types of machines that rely on different components and, therefore, different supply chains. Such restrictions could stifle innovation, increase costs and disrupt the global nature of design, testing and manufacturing processes. Limited exchange of ideas and access to new prototypes would influence the eventual nature of commercial systems and supply chains, as they did for early video cassette recorders reliant on formats such as Betamax and VHS.
Integrating quantum computers and quantum communications technologies across a coordinated network — to build a quantum internet9 — could overcome this security threat and spur growth across many industries, as the creation of the Internet did. The quantum internet is a network that connects remote quantum devices through a combination of quantum and classical links. This allows distributed quantum computing, in which many devices work together to solve problems, further speeding up computations.
The quantum internet could also enable new business models. For example, distributed quantum computers and a process known as blind quantum computing10, which allows fully private computation, could enhance machine learning while preserving proprietary data and guaranteeing that shared data are deleted after computation. Blind quantum computing would, for example, enable data or code from 3D-printing machines at a factory owned by one firm to be shared with machines at another firm’s factory without either firm seeing the details of the others’ processes. This would allow the creation and optimization of networks of factories owned by various firms to better cater to changes in product volume. Companies could offer unused 3D-printing production capacity to others, to increase efficiencies, localize production and add flexibility to supply chains.
Researchers need to determine the benefits to customers and firms of sharing data and information with faster computation, enhanced privacy, and confidentiality. Would these benefits lead to more products and services that are better tailored to customer needs? What would the impacts be on the wider industrial landscape, and what new business models might emerge?
The promise of quantum computing is great — if researchers can help to smooth the path for its implementation.
