Currently, over 20,000 Global Navigation Satellite Systems (GNSS) stations are installed worldwide to provide tropospheric delay products with high-quality and high temporal resolution (5 min). However, few of these stations are equipped with on-site meteorological sensors, leading to the challenge of the real-time transformation of this tropospheric delay into accurate precipitable water vapor (PWV). We propose a real-time high-accuracy GNSS PWV retrieval method based on artificial intelligence weather forecast foundation models. This innovative approach efficiently calculates integral zenith hydrostatic delay (ZHD) and integral weighted mean temperature (Tm) at any global location locally within seconds, bypassing inaccuracies inherent in empirical ZHD and Tm models, significantly reducing PWV errors. The method is evaluated on three representative foundation models, Pangu-Weather, GraphCast, and FengWu, showing precision loss of PWV below 0.1 mm using 6-hour forecasts. It offers a robust solution for instantaneous GNSS PWV retrieval and enhances global hazardous weather monitoring and alert systems.