Triboelectric nanogenerators (TENGs) can effectively harvest mechanical energy from the environment, offering a promising solution for a sustainable power supply in wearable electronics. However, their widespread application is often hindered by expensive raw materials and complex fabrication processes. This study develops a simple and efficient integrated ultrasonic-shear process to exfoliate low-cost flake graphite (FG) into multi-layer flake graphite (MLFG), which is then embedded into polydimethylsiloxane (PDMS) to fabricate a novel composite triboelectric layer. The multi-layered structure of MLFG provides a larger specific surface area and more charge trapping sites, significantly enhancing capacitive behavior. The optimized 2 wt% MLFG–TENG achieved an open-circuit voltage of 90.3 V and a short-circuit current of 4.6 μA, which are 1.2 times and 1.6 times higher than those of the 3 wt% FG–TENG and 3.1 times and 4.2 times higher than those of the pure PDMS–TENG, respectively. This method delivers superior output performance with lower doping levels and maintains stable output after 20 000 cycles, demonstrating exceptional scalability. Furthermore, by integrating a rectifier circuit, the MLFG–TENGs successfully power small electronic devices such as LED arrays and electronic clocks. Concurrently, when integrated with machine learning, the MLFG–TENGs achieve 100% accurate recognition of five distinct hand motion patterns, highlighting their great potential in the fields of self-powered wearable devices and motion sensing.