The author argues that Airflow, a widely used data orchestration tool, has limitations in its task-centric approach and fails to align with modern data teams' needs. In contrast, Dagster offers an asset-centric approach, focusing on data assets like tables, files, models, and notebooks, which provides a more effective way to manage complex data pipelines. The author believes that Dagster's ability to model the data ecosystem as a graph, providing lineage between actual data assets, is a significant advantage over Airflow. Additionally, Dagster offers features like rich data quality assertions, column-level lineage, cost management, and a unified data catalog, which are not yet available in Airflow 3. The author concludes that teams should consider switching to Dagster for improved productivity, observability, self-service, resource optimization, and data quality, and that the tool can integrate with existing Airflow instances, allowing for an incremental migration process.