Precision feeding is a strategy for supplying feed as close as possible to each animal’s nutrient requirements. Usually, the nutrient requirements of gestating sows are provided by a model using input data such as sow characteristics as well as an estimation of future farrowing performances. New sensors and automatons have been developed on pig farms over the last decade, generating large amounts of data. This study proposes to predict daily nutrient requirements, using data measured by sensors recorded on 73 gestating sows. Considering various digital farm configurations (for example electronic feeders and drinker stations), we explore and evaluate the performance of nine machine-learning algorithms to predict the daily metabolizable energy and standardized ileal digestible lysine requirements for each sow. Their prediction results were compared to those predicted by the InraPorc model, a mechanistic model designed for precision feeding of gestating sows. The higher correlation coefficient values for lysine (0.99) and for energy (0.95) were obtained for scenarios involving automatic feeder system alone or combined with another sensor. For the scenarios using data from the automatic feeder only, the root mean square error was lower with gradient tree boosting (0.91 MJ/d for energy and 0.08 g/d for lysine) compared to those obtained using linear regression (2.75 MJ/d and 1.07 g/d). The results of this study show that the daily nutrient requirements of gestating sows can be accurately predicted using data provided by sensors and machine-learning algorithms. This paves the way for simpler solutions in precision feeding.