The performance of electrochemical electric vehicle (EV) batteries is severely hampered by coldweather, which presents a substantial obstacle to the installation of fast-charging infrastructure as EVusage continues to rise globally. In order to challenge this difficulty, these findings suggests aprobabilistic planning approach for expanding grid-connected fast-charging stations in areas with coldclimates. EVs are coupled according to temperature-dependent charging power levels with a multi-classqueueing approach. The minimal station capacity needed to consistently service all vehicle classes isdetermined using the Loss of Load Probability (LoLP), a Quality of Service (QoS) indicator. The modelincorporates the effects of arrival rates, ambient conditions, and customer mix on station use using realworld charging and temperature data. The results shows that the proposed optimization framework canreduce the required nominal station capacity by nearly one-third compared to conventional deterministicplanning methods, while still maintaining acceptable service reliability. Furthermore, under variableoperating conditions, the suggested paradigm offers a methodical and scalable way to assess the tradeoffs between infrastructure investment and service performance. The model provides increased accuracyin forecasting effective station capacity and congestion risk by explicitly including temperature-inducedcharging limitations and stochastic demand behaviour. This makes it possible for grid operators andcharging infrastructure planners to make data-driven, well-informed decisions that strike a balancebetween capital expenditure, operational effectiveness, and user experience. All things considered, theframework is a useful tool for creating grid-integrated, climate-resilient EV charging networks that cansustain high levels of electric mobility penetration in cold climates.