The author reflects on their experience in building foundational technology for Machine Learning platforms at Apple and Uber, and how the criticality of ML data has evolved over time. They argue that organizations that obsess on the quality of their ML Data will outperform those that focus solely on models. The author highlights the evolution from Big data to ML platforms, with advancements in Machine Learning techniques, frameworks like TensorFlow and PyTorch, and the rise of Feature Stores for managing the lifecycle of ML data. However, they also note that despite these advancements, quality issues persist, including poor data quality, lack of observability, and scalability challenges. The author emphasizes the need for robust ML Data Intelligence solutions to address these challenges and ensure high-quality models are trained and evaluated on reliable data.