Proactive auto scaling can be implemented in a scaled multi-tenant cloud environment. For a particular update window, a management server can obtain tenant details for each tenant whose endpoints are to be updated during the update window using a file server. The management server can use the tenant details to calculate a total update time for deploying a respective update to each of the endpoints. If the total update time will exceed the update window, the management server can create one or more additional file servers and cause some of the endpoints to obtain their respective update from the one or more additional file servers to thereby ensure that all the endpoints can complete the update during the update window.