<p>This paper presents the first study that evaluates the effect of a carbon policy on the optimization of a friction stir welding process. Four different carbon policies, i.e., carbon tax, carbon cap, cap-and-offset, and cap-and-trade, are considered. To this end, a total of five optimization models are formulated, which include the baseline model serving as the benchmark, and four enhanced models corresponding to the four carbon policies, respectively. Each optimization model is embedded with two process input-output models built from experimental data, specifically a support vector regression model for ultimate tensile strength and a generalized linear model for weld quality (defective or non-defective), respectively. The algorithm employed to solve all optimization models is a hybrid metaheuristic, which comprises of three integrated modules: differential evolution, harmony search, and Hooke-and-Jeeves local search. A real-world load-control friction stir welding process is used as a case study to illustrate the working of each optimization model and the hybrid solution method in comparison with differential evolution. The main results indicate that the proposed hybrid metaheuristic algorithm finds better near-optimal solutions than differential evolution, by 2–20% in objective values depending upon the policy under specified parameters. Sensitivity analyses are carried out to investigate the effects of chosen model parameters. Based on the results obtained, managerial insights are highlighted, along with a discussion of relevant issues.</p>