
Marco Frezzella
Press Officer
Haus G, Max-Planck-Ring 14
98693 Ilmenau
+ 49 3677 69-5003
marco.frezzella@tu-ilmenau.de
Riyan/stock.adobe.comIn the neuroNODE research project, which starts on 1 January, researchers at TU Ilmenau are developing super-fast microelectronic components that are nevertheless extremely energy-efficient - an imperative given the rapid growth in energy requirements due to global digitalization. The superconducting circuits are inspired by human biology, which is extremely energy-efficient: they exchange information and signals in a similar way to the voltage impulses on human nerve pathways. The neuroNODE project is being funded by the Thuringian Ministry of Education, Science and Culture with 482,000 euros from the European Social Fund Plus for three years.
Everyone uses the Internet: Business and industry as well as the administration and private individuals. The use of smartphones, tablets and laptops for billions of search engine queries, followed by the emergence of streaming services, has increased energy requirements to unprecedented levels - managed in the technical background by data centers of enormous proportions. The boom in artificial intelligence has caused energy consumption to explode once again. The training of the chatbot ChatGPT 4 alone consumed an estimated 50 million kilowatt hours of energy - enough to power over 10,000 three-person households for an entire year. Current figures from the USA assume that electricity consumption for IT applications has doubled since the early days of artificial intelligence, i.e. in the last eight years. According to estimates by the Lawrence Berkeley National Laboratory, more than half of the electricity consumed by data centers will be used solely for AI by 2028.
The rapid progress of the digital revolution is in danger of being derailed by an energy demand that can no longer be met. Researchers around the world are therefore looking for alternative technologies for microelectronic circuits: They not only have to work a little, but a lot more energy-efficiently than the chips used to date.
Beate HoevelmansLeading Thuringian research teams are combining their approaches in the research project "Optoelectronic-superconducting networks for neuromorphic computing (neuroNODE)" under the leadership of Professor Hannes Töpfer, Head of the Advanced Electromagnetics Group at TU Ilmenau. Their goal: ultra-fast microelectronic components that nevertheless have extremely low electrical losses, i.e. are very energy-efficient. But that is not enough. Innovative ways are also needed to interconnect the chips into computing units in the most energy-efficient way possible and to connect them to optical modules - in short, completely new optoelectronic-superconducting networks are needed.
This is not easy because completely different material systems have to be compatible with each other in one system. Professor Töpfer's researchers will use superconducting materials because of their inherent quantum effects. Combining them with other material systems in such a way that the desired function is achieved is a challenge. For the research team, it is important to optimize all the individual components so that they work together stably and reliably.
The path taken by the research team is inspired by human nature. Superconducting circuits use various quantum effects to exchange information and signals in the form of very short pulses - just like the voltage pulses that run along human nerve pathways. Already anticipated as a vision in the 1950s by the great computer pioneer John von Neumann, the Thuringian researchers now have the technologies and the know-how to transfer this knowledge to microelectronic circuits.
The reason why the researchers are inspired by biology is simple: nature is unsurpassed in its energy efficiency. At 20 watts, the human brain uses only a fraction of the energy required by a computer to solve complex tasks. It achieves this thanks to a special feature: signals are exchanged in short pulses, so energy is not used constantly, but only when something needs to be communicated or controlled. This is exactly how the neuroNODE microelectronic components are supposed to work.
Project leader Professor Hannes Töpfer is confident that the components will not only be very fast, but also extremely energy-efficient: "Once our technology is used in data centers and then gradually improved, I estimate that we will only have to use half as much energy as we do today for the same computing power."
The development of a completely new generation of extremely powerful computers is currently being driven forward at full speed worldwide: Quantum computers. Once they are ready to function, exactly the kind of building blocks that are being researched in the neuroNODE project in Ilmenau will be needed to interconnect the multitude of components without interference. "If they don't exist," says Professor Töpfer, "it will be impossible to control the required number of quantum bits. They are therefore of existential importance for quantum computers. In this respect, our project is starting at just the right time." Technology planners in the USA assume that there will be no alternative to the use of optical superconducting electronic modules from around 2035.
Prof. Hannes Töpfer
Head of the Advanced Electromagnetics Group
+49 3677 69-2630
hannes.toepfer@tu-ilmenau.de