Over the past decades, the penetration of renewable energies has been increasing in many countries. Due to the natural intermittency of renewable energy sources, finding the optimal capacity and location of these energy resources is the most important issue in smart sustainable cities. In this regard, a mathematical strategy is proposed in this paper to find the optimal capacity and location of wind turbines and photovoltaic panels to minimize the total operation cost of a water and energy nexus. In addition, to increase the system efficiency and maximize the benefit of renewable energy sources, water desalination units have been implemented in the proposed methodology, and both electric and water networks are optimized together. In order to present a comprehensive model, the technical, environmental, and economic aspects of the problem are taken into account as a mixed-integer linear programming, where, a machine learning approach based on long-short term memory networks has been utilized for uncertainty modeling of the stochastic parameters including wind and solar generation power, electricity load demand, water demand, and electricity price. Finally, to evaluate the efficiency of the proposed method three different scenarios on a water and energy nexus have been studied considering Kuwait data. The numerical results show that by optimal planning of renewable energy sources, the total cost has decreased about $11950 that is verified the effectiveness of the proposed method. •Water and electricity nexus is considered with modeling optimal operation of units.•A techno-economic-environmental effect of different sources has been investigated.•A machine learning based method is proposed for uncertainty modeling.•An optimization framework is proposed to guarantee the global optimum results.