Nvidia has released a new version named RAPIDS cuDF, which is claimed to enable pandas to run on GPUs with a 150-fold performance improvement. RAPIDS cuDF is a Python GPU dataframe library built on Apache Arrow. With the new pandas acceleration mode, it allows unmodified pandas code to run in GPU-accelerated environments, achieving up to a 150-fold performance boost. This new version of RAPIDS cuDF addresses the limitations of using cuDF, offering a unified CPU/GPU experience, allowing pandas code to run in GPU-accelerated environments without modification. Benchmark tests show that the pandas accelerated by RAPIDS cuDF significantly outperforms pandas on CPUs in terms of task execution speed. This new feature will soon be integrated into Nvidia AI Enterprise, proving to be very useful for data scientists who wish to continue using pandas for large-scale data processing.