Energy-efficient AI for Sustainable Data Centers

Rising energy costs and the limitations of traditional hardware threaten the economic operation of German data centers, which are essential for the digital transformation with AI.  National targets for climate neutrality by 2027 and rising costs for electricity, gas, mineral oils, and coal pose a further challenge, as data centers consume an enormous amount of energy (16 billion kWh in 2020).

Energy consumption is to be reduced through neuromorphic hardware and energy-efficient AI algorithms.

ESCADE optimizes the energy efficiency of artificial intelligence in data centers. Approaches such as knowledge distillation, quantization, or neural architecture search lead to smaller, more energy-efficient AI models that nevertheless offer comparable performance. Models that can be combined with neuromorphic hardware thus enable energy savings of up to 80 %.

An ESCADE AI agent optimizes the energy efficiency of algorithms during development, training, and application. Data-based decisions enable resource-saving solutions, such as scrap sorting and software development, without sacrificing performance. SMEs, in particular, can benefit from economically and ecologically sustainable AI.