I study the optimal taxation of robots, other capital, and labor income. I show that it is optimal to distort robot adoption. The robot tax (or subsidy) exploits general-equilibrium effects to compress wages, which reduces income-tax distortions of labor supply, thereby raising welfare. In the calibrated model, when robots are expensive, a robot subsidy is optimal. As robots get cheaper, it becomes optimal to tax them. Yet, when reforming the status-quo tax system, most welfare gains can be achieved by adjusting the income tax. The additional gains from taxing robots differently than other equipment capital are close to zero.