Authors: Konstantinos Rallis, Ioannis Liliopoulos, Evangelos Tsipas, Georgios D. Varsamis, Nikolaos Melissourgos, Ioannis G. Karafyllidis, Georgios Ch. Sirakoulis, Panagiotis Dimitrakis
NCSR “Demokritos” and HPE
Quantum computing is frequently discussed as a disruptive new computing paradigm. In practice, its near-term impact is more closely linked to how it can be integrated with existing high-performance computing systems. Rather than functioning as a standalone alternative to classical machines, quantum processors are increasingly being explored as complementary resources that can operate alongside conventional computing infrastructures.
High-Performance Computing (HPC) systems already power weather forecasting, drug discovery, climate modeling, artificial intelligence and many other scientific and industrial tasks, as well as more common every-day applications. Quantum computers, on the other hand, promise breakthroughs in solving problems that quickly overwhelm even the fastest existing supercomputers. However, quantum machines cannot operate alone.
To unlock their real value and power, quantum processors must become part of a broader computing ecosystem, an ecosystem where classical and quantum resources communicate and collaborate seamlessly. This is where hybrid quantum-classical computing comes in, or as IBM refers it, quantum-centric supercomputing [1], making the physical part just as important as the algorithms themselves.
Why Quantum Needs Classical ?
Quantum computers are extremely powerful, but also extremely specialized. They excel at very specific problem categories, such as optimization, materials simulation, or certain machine-learning tasks. Everything else that surrounds the execution of a specific quantum algorithm, including input data preparation and encoding, orchestration, error handling, scheduling, and result interpretation, still relies on conventional, classical systems.
In practice, this means that quantum processors act more like accelerators, in a way similar to GPUs today. They are only called upon for the tasks they are good at, while classical systems handle the rest.
Thus, a question naturally comes in mind: How do we physically and logically connect these theoretically and structurally completely different machines?
Figure 1: A) A loose type of interconnection through the web. B) A tighter interconnection where the two systems are discrete, communicating through the web with a more complete software stack and several interconnection automations. C) Interconnections of physically sperate systems that are both located in the same area. Multiple quantum processors can be included and seamlessly operate with more advanced software stack. D) Futuristic, envisioned on-node integration of classical and quantum subsystems.
From Cloud Access to On-Node Quantum Acceleration
Today, most people interact with quantum computers through the cloud, representing a loose type of classical-quantum system interconnections. A job is sent over the network, executed remotely, and the result comes back seconds or even minutes later, depending on the task and the executed algorithm. This approach is useful for easy and fast experimentation with the system and the tested algorithms, but it introduces delays, limits real-time interaction, a topic very crucial when the actual target is maximum performance, and can also raise concerns about scalability and data security.
More advanced hybrid systems aim to reduce this distance:
– Co-located systems place quantum and classical machines in the same facility (still not in the same system), connected through high-speed networks and supported by automated mechanisms for orchestration and information exchange.
– Even more tightly integrated systems go a step further, allowing multiple quantum processors to work alongside classical nodes with minimal latency and even more efficient (and thus complicated) interconnection.
– On-node integration, the long-term vision that is still not achievable with the existing technology, embeds quantum processors directly into computing nodes, much like GPUs today.
Each step brings better performance and tighter interaction, but also increases engineering complexity. Non-trivial cooling requirements, signal integrity, and system orchestration that derive from the foundational and structural differences of quantum systems, compared to classical, become serious challenges [3].
The Role of Hardware Interfaces
Behind the aforementioned high-level interconnection approaches, and on top of the physical systems, lies the actual interfacing hardware.
This hardware-level communication between quantum processors and classical computing systems, can involve and many times combine a diverse set of technologies including [2]:
– High-speed interconnects, similar to those used for GPUs that enable low-latency data exchange.
– Dedicated controllers, often based on FPGAs or custom chips, which are responsible for translating classical instruction signals into precise quantum control signals.
– Cryogenic links, operating at extremely low temperatures that ensure quantum bits remain stable and usable.
– Emerging quantum networks, that include actually a set of different technologies which may allow quantum processors to communicate directly without the intervention of classical systems.
These interfaces play a critical role in determining how hybrid quantum-classical systems operate in practice. Their design influences system latency, data movement, and overall efficiency, directly affecting how quantum processors can be utilized within hybrid workflows.
Toward Hybrid Quantum-Classical Computing Systems
Hybrid quantum-classical computing is not a distant concept. As Europe invests in HPC, cloud infrastructures, edge computing, and AI, quantum technologies must be integrated in a way that is secure, scalable, and interoperable.
This aligns directly with the NOUS vision which involves building a coherent digital ecosystem where cloud, edge, HPC, and emerging technologies work together rather than in isolation. Within such an ecosystem, quantum processors are considered additional, yet essential resources, similar to GPUs, that can be accessed and orchestrated alongside established classical infrastructures.
The practical adoption of quantum computing is expected to depend less on isolated breakthroughs and more on the development of reliable interfaces, robust engineering practices, and system-level mechanisms that enable its controlled and effective use alongside classical computing resources, forming hybrid environments.
The future of quantum hardware will be shaped not only by increasing qubit counts or suppressing errors to extremely low levels, but also by how well quantum machines fit into existing computing systems and workflows. Standardization, low-latency communication, and hardware-software co-design will define whether quantum computing becomes a practical tool or remains restricted and confined to laboratories, only accessible by highly trained scientists.
By focusing on the interfaces and coordination mechanisms between classical and quantum systems, hybrid architectures move closer to supporting everyday high-performance workflows, while remaining accessible through already familiar computing paradigm.
References
| [1] | “What is quantum-centric supercomputing?”, https://www.ibm.com/think/topics/quantum-centric-supercomputing. |
| [2] | Rallis, Konstantinos, et al. “Interfacing Quantum Computing Systems with High-Performance Computing Systems: An Overview.” arXiv preprint arXiv:2509.06205 (2025). |
| [3] | K. Rallis et al., “Hardware-level Interfaces for Hybrid Quantum-Classical Computing Systems,” 2025 6th International Conference in Electronic Engineering & Information Technology (EEITE), Chania, Greece, 2025, pp. 1-6, doi: 10.1109/EEITE65381.2025.11166221. |
