Quantum computing is one of those technologies that futurists often joke is always a couple of decades away. It’s true that it will be a decade or two before the technology reaches maturity (in widespread use in business), but there are already opportunities for some organisations to gain early experience and benefit today. But to do that, you need a basic understanding of what quantum computing is.
Quantum computing is completely different from today’s ubiquitous digital computers. It exploits the properties of quantum physics (the physics of subatomic particles) to perform calculations and simulations that would not be practical on a non-quantum machine.
The power of the quantum computer comes from the fact that it’s not limited to binary bits (the ones and zeros of traditional computer processors). Instead, it uses quantum bits, or qubits, that can represent a one, a zero, or both at once – a ‘superposition state’ that encompasses a near infinite spectrum of probabilities.
This means quantum computers approach problems very differently – essentially attacking them concurrently rather than sequentially. They manipulate the properties of quantum entangled qubits to simultaneously try a vast number of solutions, rather than trying each in turn. This translates to dramatic increases in speed when solving certain classes of problem.
When thinking about today’s quantum computers, I find it useful to draw an analogy with the evolution of digital computers. Just as in the early days of classical computing, quantum machines are characterised by being complex, unwieldy and need to be managed carefully. They use complex mechanical engineering to keep the qubits in their entangled state.
Today’s approaches to quantum computing try to solve optimisation problems that have broad cross-sector applicability, like the fine tuning of supply chains. There are two broadly different approaches, the universal quantum computer and the adiabatic (or annealer) quantum computers such as the machines supplied by D-Wave. Rather than valves and vacuum tubes, today’s first-generation quantum computers use a variety of physical states to create qubits. These include trapped ions, superconductors, semiconductors and nitrogen vacancy diamond to stabilise qubits so they can run calculations and correct errors.
Right now, this demands significant physical infrastructure, often cooling the machine to temperatures near absolute zero and simultaneously keeping the quantum devices in a vacuum. This makes quantum computers look more like the mainframes of old than anything that would sit on a desk or fit in a pocket or bag. However, the innovation and miniaturisation that took the form fit and function of a valve in the 1950s and put over a billion transistors on silicon chips we carry around in our mobile phones is already developing in the quantum computing world.
Even with such complex construction, today’s first-generation quantum machines suffer from unstable physical qubits with limited capacity and a propensity for errors. So, engineers create logical qubits that use many physical qubits to create stability and correct errors. To be useful for business, there must be at least 49 logical qubits. Complex applications, such as simulations of real-world phenomena, will need closer to 150. The challenge is, every logical qubit potentially needs thousands of physical qubits, but the current record is Google’s 72-qubit Bristlecone processor.
Even so, current quantum computers provide a test environment for experimentation and innovation with quantum algorithms. You certainly won’t want to run your whole organisation off them but testing now will make the transition easier when quantum supremacy, the point at which a quantum computer can perform a task better than a digital computer, arrives.
Where quantum computing offers a genuine advantage over traditional approaches is when there’s a need to simulate or model complex real-world phenomena. For example, pharmaceutical drug discovery, geophysical analysis in oil and gas exploration, weather and financial forecasting, and chemical and materials science. In these sectors, some organisations are already investing in quantum capabilities, such as Volkswagen, hedge funds Renaissance and DE Shaw, and Biogen.
To gain an early advantage, organisations should begin to familiarise themselves with quantum algorithms and source the skills needed to create them. This doesn’t need significant investment in quantum computing infrastructure. IBM, for example, gives access to its quantum devices, processors and simulators via the cloud. Alternatively, organisations could seek to partner with start-ups developing quantum software, as Dow Chemical Company has with 1Qbit. Another option is to create partnerships or sponsorship arrangements with academic institutions researching the technology, as the US Army Research Office did with Yale.
As the first generation of quantum computing evolves over the next decade, a wider range of organisations will have the chance to access quantum-as-a-service capabilities via the cloud and specialised and standardised quantum algorithms. But for first movers, there are opportunities to seize today.
The financial services sector already uses serious computing power to model markets and deliver superior returns in a fiercely competitive environment. Quantum computing could run complex market simulations faster and in parallel, allowing even faster transaction optimisation.
There are opportunities for quantum computing to improve on dynamic portfolio optimisation, pricing options and derivatives, and risk management. There’s also scope to improve algorithms that rely on restrictive assumptions and heuristics imposed by the limitations of conventional machines.
Although in relative infancy today, the market for quantum computing in financial services is likely to grow rapidly as the technology matures. Early experimentation has already begun to explore how quantum computing can optimise arbitrage.
Drug development is a complex business. It needs powerful computers to model how proteins and chemicals interact at the molecular level. And with advances in genetic sequencing and the trend towards personalised treatments, there’s a growing demand to model the effects of new drugs on individuals.
The ability of quantum computers to effectively explore thousands of combinations in parallel, discarding those that don’t work, offers the opportunity to reduce the time, financial and labour costs associated with discovering new treatments.
The benefits of this kind of quantum computing go beyond drug discovery. Such modelling could help computational chemistry and materials science develop new materials, and improve the performance of existing ones, by simulating interactions at the quantum mechanical level.
Quantum computing looks to solve the kind of hard mathematical problems that conventional computers can’t. One classic example is the ‘travelling salesman problem’, where a salesmen needs to visit a number of cities in the minimum amount of time using a variety of transport. Solving this sort of optimisation challenge could significantly improve the design and efficiency of supply chain, logistics and transportation systems worldwide. Volkswagen’s 2017 experiment took less than a second to optimise taxis making the airport run in Beijing, compared to around one hour using a digital computer.
Such optimisations would let organisations respond to everything from the weather to major incidents in real time.
All organisations deal with complexity. Any software system can have millions of lines of code. Hardware can have billions of transistors. The more complex the systems, the more difficult they are to predict. How will they will perform under stress or if you make an update? Take the case of a complex real-time system which must make life or death decisions – the software that flies a plane. Quantum computing has been used in the validation and verification process of old code with a known bug which took a year to find using classical techniques. The quantum computer took a few weeks. When applied to ever more complex new planes, this could reduce the time to market by years, substantially disrupting this market.
Quantum computing’s ability to run complex simulations and parallel calculations can already improve our understanding of how systems react to a range of conditions. This could be a crucial differentiator in contingency planning today.
Despite decades of continuously improving computing power, accurately forecasting the weather is still difficult. Quantum computers have the potential to quickly process vast quantities of weather data and conduct analysis that’s too complex for classical computers. The speed of the algorithms can allow the use of real time data to influence the forecasting.
Improved weather forecasting will help all industries, especially transportation, supply chain and logistics, and agriculture. Perhaps quantum computing could even put an end to the British pastime of speculating about the weather.
These are just some of the industries quantum computing is likely to disrupt soon. In the longer term, the world could change beyond recognition.
The opportunities to gain an advantage by experimenting with quantum computers today are promising. But in the coming decades, there will be a quantum revolution. That’s because quantum’s potential far outweighs the threats and barriers to adoption.
As discussed above, we’re currently at the beginning of the quantum journey with first generation hardware. Today’s quantum computers are analogous to the early days of digital computing, where we needed large, complex machinery for basic calculations.
For this first generation of quantum computing, there are two big hardware challenges. First is the reduction of noise, which causes decoherence in quantum systems leading to loss of quantum properties. Second is improving the stability of physical qubits that perform the underlying error correction for the logical qubits running quantum calculations.
On the software side, the challenge for businesses is in creating the algorithms. Quantum algorithms are fundamentally different from classical computing algorithms and need completely different skillsets. Competition to hire those with quantum programming skills will be fierce.
The second generation of quantum computers are unlikely to arrive within the next decade. But when they do, they will also bring a more mature ‘quantum stack’ that gives organisations access to a more user-friendly and better integrated experience.
Software development kits and APIs will make it easier for all industries to use quantum computing solutions. For many, cloud-based quantum functionality will make more sense than investing in hardware, which is likely to remain mechanically challenging.
The second generation will also see tighter integration with existing technologies to provide hybrid quantum-digital solutions, with the quantum elements optimised to tackle specific classes of modelling and simulation.
The third generation of quantum computers, arriving around the mid-2030s, will see the technology continue to build a better-defined stack of ancillary services and cloud accessibility. Quantum software and algorithms will also have matured, making quantum computing applicable to a wider range of problems and attractive to a broad range of industries. Alongside the complex simulation and modelling of earlier generations, there will be new applications in areas like image search and machine learning.
What is still uncertain is whether the third generation will usher in general purpose quantum computers that have the broad range of functions of today’s digital computers. Indeed, the jury is still out on whether a general-purpose quantum machine would even be desirable. Experts speculate that the best results will come from hybrid digital-quantum machines, where the quantum processor handles certain functions in the same way graphics, vector and other specialist processors do today.
It’s also possible that improvements to digital computers could offset some of the benefits of quantum computers, driving them further into specialist applications.
As with most aspects of the future, it’s difficult to predict with a high degree of certainty exactly how the quantum computing revolution will unfold. Uncertainty is greatest around when the technology might achieve ‘quantum supremacy’ – the ability of quantum machines to solve problems that classical computers practically can’t. Some believe that it might be achieved as soon as early 2019, while others believe that for practical purposes, it may still be as much as a decade away.
Nonetheless, for some sectors, the potential opportunities presented by quantum computing are significant. Organisations would do well to start investing in understanding the technology now, experimenting with the existing first generation of quantum computers. Those that can’t see how to use the technology today should keep track of its evolution, otherwise they risk missing out on this exciting and transformative technology.