In the recent days, the Internet of Things (IoT) has assumed a significant role in automized farming applications. The farmers can receive frequent information corresponding to the soil moisture, water used, humidity, temperature, moisture, etc., by utilizing actuators and sensor nodes. Various clustering techniques used in wireless sensor network (WSN) applications have achieved energy‐efficient results. However, IoT‐enabled smart farming (SF) applications require scalable networks and long‐distance communication. The existing clustering techniques cannot provide SF solutions, and work related to communication overhead, end‐to‐end delay, latency etc., has not been taken up exhaustively. Therefore, in this article, a cross‐layer‐based hybrid particle Swarm wild horse optimizer (PSWHO) and stable routing technique (CL‐HPWSR) is designed to minimize energy consumption, communication delay, latency etc. in SF applications. Initially, the network is designed to be Bi‐concentric Hexagons. The cross‐layer‐based optimal cluster head (CH) selection mechanism is designed using the hybrid optimization algorithm‐named PSWHO to mitigate excess energy consumption problems in WSN. While clustering, cross‐layer parameters from different layers such as physical, medium access control, and network are used to compute the probability of each sensor node. Moreover, the deep learning‐based routing algorithm extended deep golden eagle neural network (EDGENN) is introduced to explore the optimal route for transmitting the data. The performance of CL‐HPWSR is implemented in Matlab by focusing energy‐efficiency, computational‐efficiency and QoS‐efficiency factors. Finally, in comparison with state‐of‐the‐art‐based SF approaches such as CL‐IoT, FEEC‐IIR, BiHCLR, PSO‐ECHS, and PSO‐C, the CL‐HPWSR has performed better in terms of energy‐efficiency, QoS‐efficiency and computational‐efficiency.