Potassium (K) is crucial for global maize (Zea mays L.) production, yet the issue of "high K fertilizer input but low utilization efficiency" in K-rich soils of Xinjiang remains underexplored. A three-year field experiment (2020, 2021, 2024) in Xinjiang evaluated the effects of reduced K application on maize growth, grain yield (GY), and K-use efficiency. Five treatments were tested: K100 (136.0 kg K2Oha-1), K60 (83.5 kg K2Oha-1), K40 (55.6 kg K2Oha-1), K0 (no K), and CK (no fertilizer). The research shows that K60 significantly outperforms K100 in terms of physiological parameters (plant height + 2.7-34.7%, leaf area index (LAI) + 6.3-26.8%, dry matter + 22.0-28.8%); GY and thousand kernel weight (TKW) improved by 6.9-15.1% and 9.3-30.3%, respectively. The potassium fertilizer productivity (PFPK) and potassium fertilizer agronomic efficiency (AEK) increased by 78-112.3% and 176.4-2085% compared to the K100. During the three-year period, the maximum net income of K60 reached 28,206 CNYha-1, which was 18.9-20.7% higher than that of K100. Regression analysis identified an optimal K rate of 82.2-85 kgha-1 for maximum yield. Least squares structural equation mode (PLS-SEM) and correlation analyses revealed that moderate K reduction enhanced vegetative growth and optimized yield structure, indirectly boosting yield, thereby directly driving net income. Thus, reducing K input can achieve "lower input with higher efficiency", offering a practical basis for optimizing K management in arid-region maize systems.