Why do we need the Superconducting Qubits and Algorithms Conference?
Dr. Jan Goetz: Quantum computing is relatively young, and an industry is gradually developing. Therefore, it is important that start-ups, tech companies, computing or data centres, as well as researchers and academia exchange ideas intensively. This is how we can accelerate the technology; the quantum advantage has not yet been proven and still needs a lot of effort. The SQA therefore offers a platform on a professional level. It's nice to see that a lot of very smart people are working together towards a common goal and I really appreciate that. Many competitors take part in the SQA and talk openly with each other. We need this exchange. Last year, the SQA took place in Helsinki, this time in Munich, next year the SQA will be in Delft, where there is also a lively development centre for quantum technologies.
The range of topics at SQA is quite broad.
Goetz: We discuss the whole computing stack here, starting with the hardware, via different layers, the quality of qubits to the software and algorithms. At this stage, we need holistic knowledge to be able to dovetail ideas and developments and improve the performance of technology.
IQM develops processors - you must be primarily interested in hardware.
Goetz: We see ourselves as a full-stack provider of quantum computers and software. But the question is what we develop ourselves and which components we integrate. IQM's core competence is clearly in processors, but we see that we can get even more performance out of co-design and partnerships, for example, when algorithms are tuned to processors and vice versa. That's why we are active in almost all areas of quantum computing; in the Munich team, we have a lot of expertise on the algorithm and application side. Especially in the early phase of a new technology, companies should build up holistic knowledge - we are not yet as far along as the semiconductor industry, which has differentiated itself.
What keeps the community busy?
Goetz: Those who focus on superconducting technologies are working through their development roadmap. That is not surprising, but it is important. Through constant progress, the ecosystem can build trust. Classic chip manufacturers present a new processor generation every twelve or 18 months, and in principle, we are already able to do that. At IQM, we started with 5 qubits on a chip; now competitors are also running processors with 20 qubits, everywhere, 54 qubits are being worked on, and processors with more than 100 qubits are being prepared. These steps show that the number of qubits as well as their quality need to be increased, and this in turn is necessary for meaningful algorithms. On the algorithm side, it is becoming apparent that mathematical tricks and coding can improve the performance of the processors. The first chips are now being developed that are strong enough for useful applications. All this makes us confident.
What advantages does superconductor technology offer compared to other methods?
Goetz: It works; we can use it to make quantum computing available and run the first applications as proofs of concept. Other technologies, such as ion traps, also work, others have not yet. The superconductor technologies are based on manufacturing methods that are known from the semiconductor industry and can be transferred very nicely to the production of qubits. These have to be cooled down very strongly, but today that is no longer a problem; there are suppliers for the necessary cooling machines. Of course, cooling needs energy, but quantum computers in the field of High-Performance Computing work with mainframe computers such as those operated by the Leibniz Computing Centre for Science. With HPC, one works in the megawatt range - a cryostat, on the other hand, usually consumes a few kilowatts and is therefore far below the power consumption of these large computer systems.
IQM Quantum Computers works closely with science and research, for example with the TUM or the Leibniz Computing Centre (LRZ), to develop quantum processors: Why?
Goetz: The cooperation is extremely beneficial. To build stronger or better quantum computers, technologies and processors have to be further developed and optimised. Knowledge transfer helps here - with academic partners we bring research results to industry and companies. We have entered into a particularly strong innovation partnership with the LRZ to develop hybrid systems in which quantum computers are integrated into HPC. We want to find out together how a quantum computer works together with a supercomputer, what interfaces are needed to integrate the technologies or how this changes software, possibly also the hardware or processors. To do this, we have to try out technologies and processes. Quantum computers have so far run largely in laboratories, where conditions are different from those in a high-performance computer centre. The partnership with the LRZ is also important because it enabled us to take our first steps onto the market - researchers will soon be able to use what we develop. The LRZ provides this access. This is pioneering work that we are doing together.
What does the future hold - will we one day use stand-alone quantum computers or does quantum computing need classical technology?
Goetz: In my opinion, it will always need both. Quantum computing is not more efficient for all of the world's problems or computing operations, but only for certain complex questions that can be mathematically classified and accelerated. Because the processors do not yet allow for large data throughput and there is still no functioning, stable memory for them, quantum computers are not yet suitable, at least not today, for tasks involving the processing and analysis of big data. For pure computing, i.e. to crunch numbers or prepare data, we need supercomputers. When the actual problems crystallise, such as more complex calculations or structures, the quantum computer can be switched on as a kind of accelerator.
So, like artificial intelligence (AI), we use quantum computing as an accelerator?
Goetz: Like AI, quantum computing offers us new computing methods that can be used to solve extremely complex computational problems better. For quantum computing, nothing is trained; it is calculated ad hoc. AI enables fundamentally different data processing methods that are based on statistics and models, with the help of which systems are trained. AI and quantum computers can work very well together. It is conceivable to have a classical computer system with integrated AI and a quantum computer next to it; parts of a problem can be solved alternately on one side or the other. In the future, we will use different computing methods side by side.
IQM stabilises 5, 20 or 54 qubits on a processor, IBM also relies on superconductor technology, already manages well over 400 qubits and wants to crack the 1000-qubit mark by the end of 2024: Is this competition too big?
Goetz: These are different approaches. IQM focuses very strongly on the quality of the qubits. Of course, you need a certain basic number of qubits for quantum computing, but if you can't produce sufficient quality in the computing operations with them, the number is useless. A processor with many qubits has no advantage if information cannot be transported without errors. If the quality is not right, most of the qubits have already expired by the time the information reaches the last qubit. That's why we focus on creating a very balanced relationship between number and quality, so that you can use the processes completely and run the algorithms that are interesting from a scientific point of view.
Nevertheless, companies from Asia and the USA seem to be further ahead in quantum computing than Germany and Europe - are we losing out again?
Goetz: We are not behind. The competition worldwide is very strong; the well-known companies from America and Asia are there. But in terms of quality and stability of the qubits, the race is far from decided; we still have to take many development steps. Speed and efficiency are also important, disciplines in which start-ups are often more agile than large companies. Speaking for IQM, we started a little later than many of our competitors; in the US, we started earlier with the development of quantum computing. But even though we started later in Europe, we are keeping up; in terms of development speed, we are at least on par, if not better. In this respect, Europe is very well positioned.
What are IQM's next plans?
Goetz: We are currently tackling the next generation of processors with 100 qubits and more, while at the same time trying to get the 54-qubit chips up and running in data centres like the LRZ and others so that the research community can benefit from them. We call these systems flagship systems, and with them, we are at the forefront of the competition. We have just launched IQM Spark, a new education system with our 5 qubits processors at an affordable price. These are intended for universities, research and training centres, on which initial proof-of-concept algorithms can run. After all, we should also provide to the quantum specialists of tomorrow.
And which companies are already buying quantum computers?
Goetz: We are currently working with companies more on the basis of application partnerships. They have their use cases, such as the simulation of physical and often also chemical processes, battery development, the development of new medicines, or optimization problems, including differential equations, which are needed everywhere in industry. Most industries now want to test quantum computers and are preparing for the technology of the future. Quantum computers are not yet being acquired on a large scale. (Interview: vs/LRZ)
Stream of SQA 2023: https://www.youtube.com/@sqaconference/streams
IQM Quantum Computers: https://www.meetiqm.com/
Quantum Computing for science: https://www.quantum.lrz.de/bits-of-qubits/detail/bringing-quantum-computing-to-scientists-work