In the context of distributed computing, scaling machine learning models refers to handling machine learning tasks at scale. This involves the ability to process and manage machine learning tasks on large datasets or across a distributed infrastructure. It includes efficient computation distribution, data parallelism, and utilizing resources across multiple nodes to address machine learning tasks on a larger scale. It's not just about adjusting hardware resources but also about designing algorithms and workflows that can effectively manage the increased complexity and volume of data in a distributed environment.