Automated Machine Learning (AutoML) systems are an important part of the machine learning engineering toolbox. However, existing solutions often require substantial amounts of compute, and therefore extended running times, to achieve competitive performance.