Cloud computing offers an attractive platform to provide resources on-demand, but currently fails to meet the corresponding latency requirements for a wide range of Internet of Things (IoT) applications. In recent years efforts have been made to distribute the cloud closer to the user environment, but they were typically limited to the fixed network infrastructure as current cloud management algorithms cannot cope with the unpredictable nature of wireless networks. This fixed deployment of clouds often does not suffice as, for some applications, the access network itself already introduces an intolerable delay in response time. Therefore we propose to extend the cloud formalism into the (wireless) IoT environment itself, incorporating the infrastructure that is already present. Given the mobile nature of local infrastructure, we refer to this as a fluid extension to the cloud, or more simply as a fluid cloud.