Rapidly increasing renewable energies bring more fluctuations to the power system and put forward higher requirement on operating reserve for maintaining the system balance. Traditional generating units are phasing out and may be insufficient to satisfy this reserve requirement in the near future. Therefore, more attentions are paid to demand-side regulation resources. Inverter air conditioners (IACs) account for about 30% of the total power consumption in cities and have huge regulation potential. However, the control of large-scale dispersed IACs is intractable. Most existing studies adopt the centralized control scheme to regulate IACs, while it requires high-cost communication infrastructure and induces privacy concerns. To address this issue, this paper investigates the distributed control scheme of large-scale dispersed IACs for providing operating reserve. Firstly, a normalization approach is proposed to uniformly quantify the regulation capacities of heterogeneous IACs. Then, a distributed consensus algorithm with nonlinear protocol is developed for IACs to achieve the regulation objective under guaranteeing customers’ comfort requirements. Based on the Lyapunov stability theorem, the convergence of the proposed algorithm on large-scale dispersed IACs is proved to be always guaranteed. Finally, numerical studies verify the feasibility and performance of the proposed method, which features better data privacy protection, less communication and computation burden than centralized control methods.