Whether in logistics, in production or for the optimal composition of investment portfolios in the financial industry: quantum computers can help companies across many industries solve complex challenges using optimization scenarios as an important entry point – if the framework conditions are right.
From the optimization of logistics planning within seconds to the rapid simulation of biochemical processes for new drugs: anyone who follows the advances in the performance of quantum computers could get the impression that this research area is all about speed. Research institutions and digital corporations routinely outdo each other in their calculations with new record values. The objective: quantum supremacy. In other words, the moment when a quantum computer can solve a task that a conventional high-performance computer would fail because it would run out of time. Spectacular, but generally irrelevant – at least in terms of business practice. Beyond the hype about the technological arms race, companies can already use quantum computing to improve specific processes. Quantum optimizers, which solve combinatorial optimization problems based on quantum annealing, are used for this.
Quantum annealing is a process in which algorithms search approximately for the global minimum of a specified objective function. Sounds abstract, but is very suitable for solving special optimization tasks. For example, it can be used to solve the backpack problem – a classic dilemma in logistics, in which a number of items of different weights and values should be stowed in a backpack in such a way as to produce the highest possible value without exceeding a certain maximum weight. This is a challenge that becomes increasingly complex as the number of variables increases.
If you replace the backpack with a large fleet of trucks and the objects with many different types of goods, you get an idea of why the processing power of a conventional computer is no longer sufficient for certain tasks. Using fundamental quantum properties such as uncertainty and tunneling, a quantum optimizer can very quickly find a solution among the maze of options that corresponds – at least approximately – to the ideal value; overall, therefore, a better solution near the theoretical optimum is found much faster.
Companies can use quantum effects not only to answer complex questions in logistics, but also to solve specific problems in production. The special characteristics of the technology help in quality assurance, for example, when the optimum pairing of components is sought in production. Many companies have to record production-related deviations, such as a bearing roll that has been drilled too deeply, as rejects. T-Systems, Telekom Innovation Laboratories (T-Labs) and OSRAM have developed together with Mix Sigma a method that allows such parts to be combined into high-quality assemblies. In the process, all components are first analyzed and stored. Next, the Mix Sigma algorithm identifies the optimal pairings and creates a flow chart – it automatically assigns the bearing roller that is drilled too deep to a bearing hub that is too long.
Ultimately, the use of quantum optimizers lends itself as an entry into the world of quantum computing to solve many economically relevant challenges: from dynamic portfolio composition in finance to the optimization of smart grid infrastructures and optimization in logistics. The prerequisite is that the task must be based on a combinatorial problem to take full advantage of the technology. This in turn requires certain framework conditions. Due to the time-limited computing time (coherence time), the required calculations must be implemented within fractions of a second and easily repeated – only through multiple calculations do quantum algorithms achieve meaningful results. It is also important to be able to mathematize the problem to be solved and fit the given number of Q bits. Furthermore, the results of the calculations always represent only an approximation of the optimal.
If the framework conditions for a use case are right, companies can already achieve amazing results based on quantum optimization. Nevertheless, the possibilities of this technology are far from exhausted. In particular, there is a lot of long-term potential in possible synergies between quantum computing and artificial intelligence (AI). However, scientists are currently still encountering major hurdles: complex AIs with deep learning capabilities often operate with millions of neurons. These are quantities that currently cannot even be approximated on a quantum computer. In addition, artificial intelligence needs a lot of data – but a quantum computers do not have the time to transfer this data, due to their short calculation intervals.
An advantage for quantum AI is that neural networks, which are widely used in AI today, are quite stable with respect to fluctuations (noise). One of the disadvantages of quantum technology is not too much of an obstacle in this case. However, this is still in the realm of pure research today In the medium term, chip technologies should also gain in importance, which may also use special quantum effects at room temperatures and accelerate AI processes selectively and not, as is usual with commercial computers, only at ultra-cold temperatures (below minus 270°C).
Although quantum computing is readily attributed a disruptive potential, it is more likely that the technology will establish itself as an important addition to the IT world in the future. While there are already many ready-to-use business cases for quantum optimization, scientists will still be busy for some time with the further development of universal quantum computing. Our goal later on will be to make the benefits of technology more accessible to businesses. Both models such as quantum infrastructure as a service (QIaaS) and quantum application as a service (Q-application-aas), together with consulting and hands-on testing around the topic, are important building blocks. What is certain is that quantum computing will become more diverse as it evolves, and we will continue to see new approaches to overcoming the previous limitations – the race for technological sovereignty has only just begun.
