Variable renewable energy is set to become a key energy source worldwide, but there is concern regarding the impact of the intermittency of its output when penetration is high. Energy system models need to tackle this issue by improving modelling resolution and scope. To allow for such modelling, more and better input datasets are needed on variable renewable energy potentials and yields. These need to be of global scope, of sufficient spatial and temporal resolution, and generated with transparent, consistent methods. This study develops the methods and applies it to generate these datasets at subnational and hourly resolution. The assessment is carried out for wind and solar technologies with consistent constraints including geographical, social and economic aspects. Features from the OpenStreetMap are converted into land cover and land use datasets and applied. Hourly energy output is simulated using NASA MERRA-2 meteorological datasets, reconciled with resource maps from the Global Wind Atlas and Global Solar Atlas platforms. Capacity supply curves are provided for 731 terrestrial zones and 339 offshore zones worldwide, along with corresponding hourly output profiles over a 10-year simulation period. The proposed energy potentials are relative conservative compared with other studies. The datasets can serve as input for regional or global energy system models when analyzing high variable renewable energy shares.