According to Sawyer Merritt on X, Tesla is deploying an updated machine learning model that predicts a vehicle’s intent to charge using 9 million miles of aggregated and anonymized vehicle trajectory data collected within Supercharger geofences, aiming to reduce queue lengths and idle time at sites. As reported by Sawyer Merritt, the model improves demand forecasting and dynamic load balancing so site availability and routing in the in-car navigator can be optimized in real time, which can increase charger utilization and lower operational bottlenecks for Tesla’s charging business. According to Sawyer Merritt, better intent prediction also helps prioritize vehicles likely to plug in soon, informing smarter congestion control and potential preconditioning guidance that shortens charge sessions and improves throughput.