Winding processes are known from the fiber composite industry for strength and weight optimized lightweight components. To achieve high resistance and low weight, mainly synthetic materials are used such as carbon or glass fibers, bonded with petrochemical matrices. For the construction industry, these additive processes present a very promising and resource-efficient building technology, yet they are still hardly used with sustainable materials such as natural fibers or timber.The 3DWoodWind research prototype has developed a new generation of additive technologies to wood construction. The modular building system is built with a three-dimensional robotic winding process for material-efficient hollow lightweight components. An AI-controlled design logic enables the intelligent combination and design of modular components into multi-story structures, which may be used in the future to substitute solid wood panels and beams as well as concrete slabs and steel sections.Our current research uses a continuous strip of thin timber veneer, which is a waste product from the plywood industry and therefore, presents a highly sustainable alternative to synthetic fibers usually used in winding, as well as solid timber products known in construction. The veneer’s natural fibers are intact and continuous, and offer high tensile strength. In the presented project, three-dimensional winding processes were developed for material-efficient lightweight components made of wood. The demonstrator presents a modular column and ceiling system, which aims at large scale applications in multi-level structures. Having won an open national design competition for Germany’s ‘ZukunftBau’ Pavilion, a first demonstrator is currently being built to be presented in May 2022, as part of the DigitalBau exhibition. The paper discusses all planning engineering and production processes in detail with particular emphasis on the machine-learning algorithm, which was trained during the design process to facilitate design iterations and future planning with this component-based building system.KeywordsAdditive manufacturingWindingFE-modelingMachine learning