A workload adaptation service of a provider network may perform smart deployment of industrial an IoT workload across resources of a provider network and a client network. The workload adaptation service may receive, from a user, an indication of one or more constraints for performance of a workload (e.g., daily upload limit, maximum acceptable latency). The service determines a deployment for the workload across client resources and provider resources based at least on the constraints. The service deploys portions of the workload to the client network and other portions to the provider network according to the deployment. The workload adaptation service may also perform dynamic adaptation of the IoT workload by moving portions of the workload from the client network to the provider network or vice-versa, based on workload performance metrics or based on changes to the available resources.