Growth of data centers and their capacities is central to economies as digital transformation needs computing power. But processing data and training large AI models generates tons of CO2 and consumes a lot of energy: 16 billion kWh in 2020 with increasing trend. Rising energy costs and the requirement for climate neutrality of national data centers by 2027 threaten the ability of German economy to operate data centers economically. This poses the challenge for data centers to reduce resource consumption while maintaining the same computing power, which is impossible with existing hardware (CPU/GPU). The goal of ESCADE is to significantly improve the sustainability balance of AI applications in data centers. For this purpose, an ecosystem for AI use cases with proven sustainability is to be developed, which are operated by innovative AI services in sustainability-optimized data centers (primarily by using neuromorphic chip technologies as a supplement and replacement for traditional data center hardware). Thus, ESCADE develops end-to-end solutions that are completely based on national components and competencies in order to create jobs and strengthen the German economy. By application in the high-energy steel industry and the energy-intensive software industry, ESCADE also aims to demonstrate significant industry-specific energy-saving potential.
Partners
Deutsches Forschungszentrum für Künstliche Intelligenz (DFKI) GmbH (Koordination), NT Neue Technologie AG (NT.AG), die Stahl-Holding-Saar GmbH & Co. KGaA (SHS), die SEITEC GmbH (Seitec) sowie die Technische Universität Dresden (TUD) , die Ruhr-Universität Bochum (RUB) sowie der österreichische Partner Salzburg Research Forschungsgesellschaft m.b.H.
Außerdem die Unterauftragnehmer eco2050 Institut für Nachhaltigkeit – Institute for Sustainability GmbH (eco2050), SpiNNcloud Systems GmbH (SpiNN), elevait GmbH & Co. KG (elevait) und Aleph Alpha GmbH (AA)
