Why are telecoms interested in quantum computing?
As networks become more interconnected, critical and dynamic, we are reaching the boundaries of what classical computing can do: today carriers need to make decisions in real time in ever more complex contexts. So-called combinatorial optimisation models make it possible to automatically determine the best decisions and actions for networks thanks to calculations performed by computers. However, if we continued to address these problems using only traditional computing approaches, the amount of time and energy invested in such calculations would soon become unsustainable.
This is where quantum computing comes in (see volumes one and two of “Introduction à l’informatique quantique” [introduction to quantum computing]. For more details, go to https://perso.isima.fr/~lacomme/site3/livres.html). While classical computing is limited to exploring scenarios one by one, quantum computing enables a large number of scenarios to be explored simultaneously. It is therefore expected to unlock substantial gains in terms of performance and energy consumption in key areas such as combinatorial optimisation applied to network operations.
Quantum computing will make it possible to cross computational boundaries that up until this point have been unattainable using classical computing in telecom networks. This in turn will have a great impact on customer experience and the environment.
The future of telecommunications will therefore likely require the use of quantum computing to dynamically manage networks, improve performance and reduce energy consumption.
Optimising networks with quantum computing
Optimisation problems that have been a focus of study in telecoms include optimising resilience, routing and service orchestration in networks. The most suitable models to solve these problems are “operations research” models. Operations research is a discipline aimed at studying complex decision/optimisation problems. The challenges are significant: the task is to identify potential sources of critical failures, or to determine both the placement of network computing functions, by deploying them on cloud resources, and the routing of traffic flows from these services in order to optimise service quality criteria, such as latency. These combinatorial problems are solved by algorithms that when executed on classical computers require considerable computing resources and a long time to resolve the problem, thereby consuming a lot of energy. Orange’s research on the subject shows that it is possible to drastically reduce the computing time and energy required, while improving the service quality, by combining quantum computing with classical approaches. This is called a “hybrid model” because one part of the process is classical and the other part is quantum. Thus, we can now envisage the development of new approaches and algorithms for managing dynamic and reliable networks, providing quality solutions while reducing the amount of computing power required and the carrier’s carbon footprint.
These approaches can be adapted to resolve many other optimisation problems. Orange’s quantum computing research will therefore ultimately enable optimisation approaches to be developed and mastered in many sectors of tech.
Proactive failure management: quantum computing supports network resilience
In terms of resilience, detecting critical nodes in a network is crucial to decide on how to best allocate resources to proactively manage failures.
Quantum algorithms require less time to identify the set of nodes which, if they were to fail simultaneously, would notably impair certain services. If the probability of the impairment crosses a certain threshold, the technical teams then take action on said nodes to make them more resistant to failures.
This is where quantum annealing comes into play. This type of quantum computing is based on an existing process in classical computing that the Orange Research teams have already mastered. The premise is to search for the global minimum of a cost function (like in classical computing) but to do so using more quantum phenomena such as tunnelling (the ability to develop a cost function with constant energy), which makes it possible to boost performance. Unlike classical algorithms, which may get stuck in local minima, quantum annealing can pass through these barriers and identify optimal solutions more quickly.
This type of approach means billions of network configurations can be explored and the best-performing configuration or migration scenarios can be identified more quickly, based on criteria such as latency, resilience or cost.
Orange Research’s first results on the subject show a significant improvement in the robustness of dynamic defence strategies against failures [https://roadef2025.org/wp-content/uploads/2025/02/ROADEF2025_resumes.pdf, page 22 of the pdf]. The results provided by the hybrid model show that the risk of losing service on a part of the network can be halved in the event of several nodes failing simultaneously.
A step towards enhanced digital twins
Another promising avenue is quantum digital twins. Today, simulations are typically run using digital twins. By simulating entire networks via digital environments and quantum algorithms, complex scenarios can be explored while integrating uncertainty, unexpected incidents or non-deterministic behaviours. This paves the way for enhanced prediction capabilities and better tools to support decision-making, especially in partially observable environments.
Quantum algorithms such as Grover’s algorithm could be used efficiently, and provide a very clear advantage compared to classical approaches alone, to find optimal solutions in exponentially large spaces that can only be explored by simulation. However, such an advantage can only be observed with a sufficiently large number of qubits, probably several hundred or even thousands of “logical qubits”. These are qubits that can be directly used for computations. This is far from the case at the moment (only a few dozen logical qubits are available at best). Some players in the field, particularly in France, are already standing out in the race for quantum technology (Pasqal, Quandela, Alice&Bob), in a context where Europe is looking to become the “quantum valley” of the world (European declaration on quantum technologies signed by 11 EU Member States, including France).
An overview of superconducting quantum technologies (capable of rotating Grover natively) is available here: https://link.springer.com/content/pdf/10.1140/epja/s10050-023-01006-7.pdf
When classical meets quantum: a strategic alliance
Rather than completely replacing the existing system, quantum computing adopts a hybrid approach. By combining it with the best classical operations research algorithms (like the technology proposed by D-Wave for example, via a local search method) or artificial intelligence (as proposed by Quandela, for example, with a gradient descent method), the exploration of options and convergence on optimal solutions can be sped up.
What are the intended use cases? Intelligent autonomous systems to manage networks, such as network orchestrators.
A strategic investment for Orange
Orange is actively engaging its research teams to explore these opportunities. In 2025 a research project on quantum computing was launched, including a CIFRE (Conventions Industrielles de Formation par la Recherche — Industrial Agreements for Training through Research) thesis.
