You work for a rapidly growing social media company that handles a high volume of data, including billions of historical user events and 100,000 categorical features. Currently, your team builds TensorFlow recommender models using an on-premises CPU cluster. As the dataset continues to grow, you observe a significant increase in model training time. To address this, you are considering migrating the models to Google Cloud to leverage its scalable infrastructure and minimize training time. What approach should you adopt to achieve this goal?