<p>Energy-efficient routing in large-scale Intelligent Internet of Things (IIoT) systems incorporated with Software-Defined Networks (SDN) poses an essential challenge due to resource limitations, dynamic network environments, and balancing scalability, energy efficiency, and reliable data transmission. Thus, the proposed study presents a routing framework named Sculptor Dream Optimization with Heterogeneous Deep Maxout Graph Convolutional Neural Network (SDO-HDMGCNN), which is designed to optimize intelligent IoT networks. The suggested solution makes use of the Fuzzy C-Means clustering technique to streamline the load balancing of dynamism, and a cluster table facilitates the management of the cluster heads and client nodes. The sculptor dream optimization algorithm is the combination of the sculptor optimization of cluster head selection with the dream optimization of routing to trade off the allocation of energy and the optimum path selections. Furthermore, an overall heterogeneous deep maxout graph convolutional network is used to optimistically position mobile sink paths so as to use less energy and enhance the efficiency of data collection. This is achieved by introducing an improved process of energy balance to achieve a balanced distribution of energy used on the network, thus extending the network lifetime and preventing premature failure of nodes. The performance analysis shows that the suggested mechanism attains a network lifetime of 1900s, a throughput of 10 Mbps, and a delay of 16&#xa0;ms, which is much better than other mechanisms. This illustrates the efficiency of the suggested solution in improving the energy efficiency, scalability, and reliability in energy-efficient routing.</p>