Is quantum computing the next digital hype? Will it be disruptive? How will it be used? To answer these and more questions, we interviewed HPC expert (High Performance Computing) Huub Stoffers at his office at the Amsterdam Science Park.

Amsterdam, 20 February 2025 – Quantum computing is one of the most recent digital innovations the EU is exploring. We want to understand its application domains, its strategic positioning in the European digital technology landscape, its relation to HPC and maybe to Artificial Intelligence (AI), its future outlook, also given the hooks to industry and Big Tech in the technological race between the EU, US and China.

Huub, as a senior HPC expert, contributed to the definition of the system requirements for the national supercomputer in the Netherlands since 2000, as this is commissioned by NWO = Nationaal Groeifonds (the National Science Foundation in the Netherlands). The national academic supercomputers in the Netherlands usually have a phased deployment and average life-time of five to six years. Supercomputers have been available to the Dutch academic community since the late 1980s, always at the Science Park in Amsterdam (formally at SARA, which later became part of SURF). Huub has recently been involved in the design of the project proposal to tender for a quantum computer to be integrated with the national supercomputer. Quantum computing is one of the key technologies in the EuroStack European strategy, and relevant for the ADS-DEP programme.

Anna: Dear Huub, we have read in the Dutch newspaper shortly, that the Netherlands is going to host a new academic quantum computer at the national academic HPC centre SURF. Can you tell us a bit more about quantum computers? Is this the next hype after AI? What can we expect from it and how are you involved in this project?

Huub: I have been involved in the preparation of a project proposal submitted to EuroHPC for the hosting of a national quantum computer in the Netherlands, at the academic computer centre in Amsterdam (SURF), with the aim to embed it in the existing national supercomputing infrastructure.

EuroHPC, on behalf of the European Commission, wants to stimulate quantum computing, given the fact and ambition that this is a technology in which Europe may take the lead in the world competition, beating even US and China. We already have a lot of expertise in the field of HPC in Europe. EuroHPC wants to stimulate this digital innovation by taking quantum computing more out of the domain of quantum physics research, and more into application-oriented research, by exploring opportunities in possible application domains from the wider HPC fields such as climate modelling, fluid dynamics, aerodynamics, thermodynamics, astrophysics etc.

Currently the knowledge of quantum computing and quantum networking itself is increasing by the day. What is absolutely lacking is an Application Programming Interface (API), which is something needed to make it available on a larger scale for computational problem solving. Ideally, the experimenting with a quantum computer will transform it from a piece of hardware into a “device” – a more or less standardized abstraction of the hardware with a protocol to offload computation to and get results back.

The idea now is to host this new national quantum computer at the same datacenter as the current Dutch national academic supercomputer Snellius, so in Amsterdam, the Sience Park. We will make make it available for our HPC users to run, test, optimize their research programs on it and we will learn from it.

What does that mean for HPC? A few years ago we had the rapid evolution of GPUs. The G stands for Graphics. These have shown to be very useful not only for graphics but for many other computing applications. GPUs have therefore been massively integrated into HPC infrastructures. Although that might be a bit more complicated and take longer, the same is expected to happen with quantum devices.

Since many years we have had in the Netherlands a growing number of academic HPC users from chemistry, physics and various other computation intensive research domains who are keen to increase the processing capacity of HPC. This will allow them to improve their models, make them larger, better, higher resolution, more accuracy, greater detail. So it will bring them new avenues to explore their fields in depth. This will lead to new breakthroughs. This is what we expect with quantum.

The interface with the HPC system is what we mainly be working on. This is a large scale initiative for the whole of Europe. EuroHPC aims at establishing as many different application domains of quantum as possible in countries in Europe. We will learn most from it by having as heterogeneous and diverse implementations of quantum as possible. We want to know what the best way of integration is. What do we find out out to be difficult? What are the challenges? What are the best practices?

Anna: Which is type of hardware will there be hosted here in Amsterdam?

Huub: It will be a quantum chip in silicon, actually the same material as wat is currently used in CPUs and GPUs. It is a technology in which for example researchers at TU Delft have gained much expertise at an early stage. Ronald Hanson is one of the leading reseachers in quantum physics, also world wide. TU Delft have worked on qbits and have made important breakthroughs in 2006, 2008 and again in 2014. But this is research on qbits from the quantum physical perspective.

For us computing people, the qbit represents the basic unit of information in the quantum computer. On a limited scale it is possible to simulate qbits and a quantum computer with a few qbits on a conventional supercomputer. At SURF we have done that too. But simulated qbits are logical qbits, which are stable. With real qbits stability is a major technological issue. For one thing, to function properly, they must be kept at temperatures close to absolute zero temperature, that is: close to 0 degrees Kelvin, which is slightly lower than −273 degrees Celcius. This is solved by having multiple level of cooling. The quantum chip is very small, but it is in highly sophisticated fridge. So cooling technology, “cryogenic” technology, is one important aspect. But since you also also want to use the thing, there is more to it than the cooling per se. To use it, you have to get signals in and signals out, and every line in and line out is of course a risk of heat leaking in.

Anna: Will this technology consume a lot of energy for that reason?

Huub: On the contrary, quantum hardware is very energy efficient. A rack of CPU-only or GPU-enhanced compute nodes, and also the cooling of a such a rack in a datacentre, especially when that is still chiefly done by blowing cold a over hot silicon, consumes a lot more energy than a quantum device and its cooling. So, when integrated into the supercomputer infrastructure, the combination as a whole is more energy efficient than just using the more conventional hardware. The cooling of the quantum device requires
helium and nitrogen in the datacentre. To store and handle that correctly and safely introduces more procedural complexity for the datacentre.

Anna: How do you see the use of quantum computing in relation to AI?

Huub: That is at this moment not yet very easy to predict. We are at an early stage of development, looking into hardware integration first.

Anna: In this ongoing trend of exponentially growing computing power of increasing processing speed, increasing data storage capacity and increasingly widening network bandwidth, as predicted by Gordon Moore in the 1950s, and which has continued ever since until present day for conventional computing, where in this constellation do we place quantum? Is quantum computing going to be disruptive in this sense?

Huub: If you look at “Moore’s law”, which of course is not a physical law, CPU technology hasn´t been able to keep up with it in the last decade. But sure, we hope something
disruptive, in that sense, will happen. However, I am also a bit of a quantum sceptic. I see that a lot of issues still have to be solved technologically. The technology is far from mature before it will be stable. Qbit stability and error correction are still a problem. I am not a quantum expert, but the way it looks at present,
the number of extra qbits you need for error corrections grows exponentially. Yet, commercial parties tend to de-emphasize that aspect for the general public.
But we hope that these things will all be solved in the near future.

Anna: How is the data processing of quantum?

Huub: The difference with conventional computing is the fact that the unit is not as in a conventional bit, which holds the state of the smallest unit in which data is represented, can be either O or 1, nothing else. The qbit on the contrary can hold many different states at the same time. We do not know what the consequences will be for computing and which algorithms are suitable for quantum and which or not.

Anna: What we see with Machine Learning is that a lot of algorithms are black boxes in which the outcomes of calculations are not explainable for humans. Will that also be the case for quantum computing?

Huub: I don’t think that will make any difference. We know how algorithms work in concept. They consist in the basis of statements, variables and loops and calculations carried out on matrices of data. In the case of very large datasets such as LLMs the uncertainty is inherent of the statistical methods used for the calculation. But that will not be different.

In quantum computing we are a step lower in the conceptual stack. And currently we are still in the phase of testing out. Of course in AI research many experiments are going on to improve its performance through hardware. An example is neuromorphic computing. This is a new computer paradigm inspired by the structure and function of the human brain. It aims to create computer systems that mimic the brain’s neural networks, using artificial neurons and synapses to process information. There are currently many new developments and areas of experimentation in specific hardware for AI. Yet, currently GPU is still the most widely used hardware for AI. What that will be in a few years is hard to predict.

At this moment in our centre AI and HPC are not yet very integrated, although AI jobs are run on the national supercomputer, these are just a number of academic projects
that are run on fairly general purpose machine that has sufficient GPU-nodes to accommodate them. We are also working on a partly EuroHPC funded AI factory. But that is
completely separate from the quantum project. The AI factory, if it comes about, will definitely be a separate system with its own community of end-users, that will overlap to some extent with Dutch academia, but will also have other stake holders. From a hardware perspective, the AI factory will invest heavily in GPU-capacity for computation, and in
storage technologies because AI model training tends to be very data-intensive. Quantum technology however, is not envisaged to play any role there at all.

Anna: Do you think that quantum, started now as an academic endeavour, will soon be taken over by industry, of course I am referring to Big Tech. Will it happen similar as with the Internet and many other digital technologies in the past?

Huub: Now you are asking me to ventilate a personal opinion in the political sphere. My ideas and orientation towards society are inspired by the social democrats, in the Netherlands the political school of thought of Joop den Uyl. The motto of his cabinet was the more equal spread of knowledge, power, and income. I strongly believe that applications of technology must be more strictly regulated by the government, either on a national level, or, more effectively, on a regional level as the EU. And not only regulated. Adopted regulation must be eforced, following the well established rule of law. That is not always happening, but this current omission in the legal frameworks should not stand in the way of reserach for new future technological developments. I think that the concern for misuse of technology by industry should not withhold the academic world of doing their scientific research and development work.

If fear of misuse would be a guiding principle for us, we might stop doing any scientific or technological research at all There is always a risk of future misuse by certain groups, this holds for any new technological development. But I hope Europe will do their best to regulate digital technologies properly. It should not be the case that benefits of the new technology should go to a few corporate businesses only. Do you know how many new technology applications have been cut off for example by Google in recent years? There is an impressive list on a website “Killed by Google”, who are trying to keep track of this. Many technologies that had a great promise for the future and the outlook for great application domains in society have simply not made it to the world, because they have been seized and stopped by Big Tech. In that sense we don’t know the future of quantum. I am in favour of value-free research. Although there is a positive or negative outlook for a certain technology, should we then stop exploring? That would prevent many flowers to bloom.

Anna: OK Huub, that is a very clear statement. Let’s make this into an interesting conclusion to our very enriching conversation. Many thanks Huub, we will continue our discussions on technology and quantum, hopefully soon! Thanks on behalf of the EURIDICE team.

With the topic of quantum computing in mind, in the light of our EURIDICE ecosystem, Stichting SURF is one of the possible partnerships with whom we explore possible internships with our European EURIDICE DigiSoc students.