This work compares the power and energy used by a robotic linkage actuated by either joint motors or scissored pairs of control moment gyroscopes. The objectives are to establish straightforward sizing equations that provide a basis for deciding on a system architecture and to validate them with detailed models. The resulting parallels between joint motor and control moment gyroscope actuation increase intuition for and inform the design of control moment gyroscopes on a single-body satellite. Control moment gyroscopes are chosen as an energy-efficient means of reactionless actuation that reduce nonlinearities and coupling between the robot and the spacecraft attitude control system. Scissored-pair control moment gyroscopes are well suited for robotics applications because the output torque acts only along the joint axis, eliminating undesirable gyroscopic reaction torques. Both analysis and simulation of a single-link robot demonstrate that the control moment gyroscope power is equal to the equivalent joint motor power for a large range of gimbal inertias and maximum gimbal angles. The transverse rate of the link does not affect this result. A two-link robot with orthogonal joint axes gives results similar to the single-link system unless momentum is not conserved about the joint. For a two-link robot with parallel joint axes, control moment gyroscopes outperform joint motors in power required when the joints rotate with opposite sign; the reverse is true when the joints act in unison. These surprising differences arise because control moment gyroscopes produce body torques with a zerotorque boundary condition at the joint, whereas joint motors produce torques that are reacted onto two adjacent links. The analysis concludes with pros and cons of control moment gyroscopes as robotic joint actuators.