Depends on application, companies in which ML is part of the main product will be more robust, experimental/research/marketing will probably see more issues, data scientists who aren't already operational with proven numbers might take a large hit