We demonstrate a 40-Gb/s PAM-4 transmitter having 2-tap pre-emphasis, whose power consumption is significantly reduced by the use of toggling serializers. In addition, a new type of PAM-4 source-series terminated driver is proposed, which enables an easy control of pre-emphasis gain while maintaining impedance matching. A prototype 40-Gb/s PAM-4 transmitter containing these circuit ideas are successfully realized in 28-nm CMOS technology. It occupies 0.006 mm2 and achieves energy efficiencies of 1.2 pJ/bit for 40-Gb/s operation without any pre-emphasis, and 1.68 pJ/bit with 9.54 dB pre-emphasis gain.