Synthetic biology’s remarkable potential to tackle important societal problems is held back by our inability to predictably engineer biological systems. Here, we collected one of the largest public multiomics synthetic biology datasets generated to date, and used it to train machine learning algorithms that are able to predict product and metabolic dynamics with great accuracy, starting to approach the predictive capabilities found in physics and chemistry. These predictions were provided for a non-model yeast, Pichia kudriavzevii, displaying carbon fluxes very different from model yeasts, and already engineered to produce large amounts of malonic acid, a desirable biomanufacturing target. Recommendations for genetic interventions obtained from this model improved final production, and illustrated existing limitations in leveraging predictive models of metabolism. This work shows the key role that machine learning can potentially play to significantly reduce the development time of novel synthetic biology products, paving the way for a timely replacement of the thousands of petrochemicals used in the modern economy, to create a working circular bioeconomy that can help mitigate climate change.