The proposal applies for extended storage space for NAISS 2025/5-398 to be able to install & run protein design tools locally (including AlphaFold3, ProteinMPNN, RfDiffusion, MD trajectories etc). This project is funded by VR and Novo Nordisk Foundation individual grants to Per-Olof Syrén as well as by SSF, MISTRA, FORMAS. Published papers using NAISS 2024/23-435 & NAISS 2024/5-346 4, and 1 manuscript to be submitted. This project will use de novo enzyme design to bridge transition metal catalysis with protein chemistries for the generation of new-to-nature enzymatic reactions. Enzymes show a great potential towards reaching a more sustainable society by allowing for specific transformations starting even from complex biomass under mild conditions. However, for many industrially important chemical reactions, an enzyme is lacking. This is particularly true for carbon-carbon and carbon-heteroatom bond formation which is of basal importance to make platform chemicals, materials, agrochemicals and pharmaceuticals. We are thus still today often limited to harsh chemical synthesis relying on petroleum-based feedstock and hazardous conditions. Here, we will create metalloenzymes de novo to create carbon-carbon bond forming enzymes with broad catalytic scope. Using computational methods spanning from quantum mechanics (QM), molecular dynamics (MD) and generative AI-based protein design we will fill current gaps in available biocatalytic transformations. We are focusing on cross-coupling reactions for which our devised pipeline of protein design consists of: 1) QM-calculation of a transition state for the relevant reaction (here cross-couplings) 2) GPU-based generative AI to diffuse a protein backbone capable of harboring this reaction 3) Validation by AlphaFold followed by MD and docking to find hot-spots amenable for further engineering. Designs are tested in the lab experimentally and further engineered by directed evolution to reach efficient catalytic rates. By implementing new reactions and mechanisms in de novo metalloenzymes guided by high-level QM calculations and experimental data, the result are designer enzymes capable of harnessing chemical transformations of high industrial and societal relevance that are currently not accessible. The PI has experience in modelling enzymes and great progress was made in previous NAISS supported projects, as we unraveled new chemistries in existing proteins by starting building and testing the proposed pipeline (Syrén et al. J. Green. Chem. & Methods. Enz. 2024). When the now ready-to-use pipeline above will be implemented in NAISS 2025/5-398, the necessity to have storage will increase (e.g. to store both results and model weights from NVIDIA GPU:s). Motivation to Multicenter usage: Dardel: Per-Olof Syrén (PI) has several funded projects from VR, FORMAS, MISTRA and SSF whose success are dependent on running US-GAMESS. Calculations are based on very large models of complex biocatalysts containing metals. Tetralith: This project is dependent on GPUs to run molecular dynamics (openMM) and generative AI for protein structure prediction (AlphaFold) and design (RfDiffussonAA etc). We also use Schrödinger and Orca. Ca 6 people from my group will use these resources