Apparatuses, methods, systems, and program products are disclosed for data packet prediction. An apparatus includes a processor and a memory that stores code executable by the processor. The code is executable by the processor to train a predictive model using incoming data packets. The incoming data packets are forwarded to one or more network devices. The predictive model is trained to predict a subsequent data packet based on an input data packet. The code is executable by the processor to forward the predictive model to the one or more network devices in response to the predictive accuracy of the predictive model satisfying a threshold accuracy. The predictive model generates predicted data packets at the one or more network devices. The code is executable by the processor to cease forwarding incoming data packets to the one or more network devices.