This storage application is linked to the GPU part of the "S-CMIP: Swedish Climate Modelling and contributions to International Projects" - proposal. The warming of the Earth continues with unprecedented speed, exceeding in 2025 1.5 Celsius compared to pre-industrial value indicating the urgency to improve understanding of the risks and consequences related to the ongoing climate change. The goal of S-CMIP is to better describe and understand the Earth system, its response to changes in greenhouse gas forcing and land-use, its internal variability and the interactions between Earth system components. Scientific progress in these areas will enable actionable information on climate change in the fields of climate change adaptation (adjustment to a new climate) and mitigation (control of greenhouse gas emissions). In addition to traditional dynamical models, which provide the core of S-CMIP, machine learning (ML) methods become increasingly relevant in the development and interpretation of climate models and model simulations. We, thus, for the first time include ML-applications to support and complement traditional dynamical simulations and analysis. The focus will be mainly on regional downscaling, detection of extremes, generation of ensemble members, and emulation of high-resolution processes. The work will be performed as part of the European Horizon projects OptimESM, AI4PEX and SEACLIM, as well as part of the Destination Earth project DEODE.