This paper proposes a real-time energy management strategy (EMS) for fuel cell hybrid electric trucks (FCHETs). In such trucks, hydrogen fuel cells (FCs) are employed as the primary power source while lithium-ion batteries can be used as the energy buffer. Since hydrogen is still expensive and the FC technology readiness level (TRL) is still low, properly-designed EMSs are effective approaches for boosting the fuel economy of FCHETs. Therefore, an EMS is proposed in this paper to determine the optimal power allocation between the FC and the battery in FCHETs. The EMS is composed of three blocks: alternating direction method of multipliers (ADMM), rolling optimization, and speed prediction. The EMS is implemented on the basis of a prediction horizon. For a particular horizon, the speed prediction block predicts the FCHET speed using a long short-term memory (LSTM) network. Based on the predicted speed, the electric machine power is determined, which is further split and allocated to the FC and the battery using the ADMM algorithm. The effectiveness of the EMS is demonstrated by performing real-time simulations using a real-time system based on a dSPACE DS1202 platform. Simulation results show that the EMS can effectively split and allocate the load power to minimize the hydrogen cost and balance the battery energy by reducing the hydrogen consumption and maintaining the battery state of charge (SOC), respectively. Moreover, the EMS is evaluated against a CVX-based framework. Evaluation results show that compared to the CVX-based framework, the EMS leads to 5.94% and 14.31% lower hydrogen costs as well as 85.58% and 95.66% lower battery SOC changes for the modified CYC_WVUCITY and CYC_NYCTRUCK drive cycles, respectively.