Developing ETL pipelines using Databricks comes with several key challenges:

1. Data Quality and Consistency: Handling messy, inconsistent, or incomplete data can affect pipeline reliability. Ensuring proper validation and cleaning is essential.

2. Performance Optimization: Processing large datasets efficiently requires careful tuning of Spark jobs, cluster configuration, and partitioning strategies.

3. Complex Transformations: Implementing complex business logic or aggregations in distributed environments can be tricky and may require advanced Spark knowledge.

4. Scalability and Resource Management: Ensuring pipelines scale with growing data volumes while keeping costs under control is a common concern.

5. Error Handling and Monitoring: Building robust pipelines requires logging, error recovery mechanisms, and monitoring for failures.

6. Integration with External Systems: Connecting Databricks pipelines with databases, cloud storage, and third-party tools can introduce compatibility and latency issues.

7. Version Control and Collaboration: Managing code versions and enabling team collaboration within Databricks notebooks can be challenging without proper practices.

By addressing these challenges through best practices, automation, and testing, ETL pipelines in Databricks can become efficient, reliable, and maintainable.