AWS Cuts Glue 6.0 Pricing by 30% and Backs Iceberg v3 to Slash Serverless ETL TCO

Amazon Web Services has released AWS Glue 6.0, introducing a direct 30% price cut on standard compute usage rates alongside modernized runtime engines including Apache Spark 4.1.1, Python 3.13, and Scala 2.13. In addition to the baseline rate cut, the release delivers complete support for the Apache Iceberg v3 specification, featuring native VARIANT data types with automatic column shredding, deletion vectors for granular row-level updates, and nanosecond timestamp resolution. The update also adds Spark Declarative Pipelines and a sub-second Real-Time streaming execution mode. Serverless data processing on AWS has historically carried a price premium compared to self-managed Amazon EMR or Kubernetes-based Spark clusters. By combining a 30% lower rate card with runtime optimizations that deliver up to 36% better price-performance on benchmarked workloads, Glue 6.0 significantly compresses the total cost of ownership (TCO) for enterprise ETL pipelines. The native VARIANT support in Iceberg v3 eliminates the need to maintain redundant flattened tables or write compute-heavy custom JSON parsing logic, directly tackling query and storage cost inflation in high-volume logging and telemetry pipelines. This move highlights the ongoing convergence of FinOps principles and open data lakehouse architectures. As enterprises face growing pressure to rein in analytics and AI data pipeline expenditures, hyperscalers are being forced to offer aggressive compute efficiency improvements and open table format interoperability. Embracing Iceberg v3 alongside price cuts allows AWS to defend its serverless processing market against external ETL SaaS vendors and lakehouse platforms, keeping data workloads within native cloud boundaries while maintaining workload portability. Practitioners should audit existing Glue jobs for immediate migration to the 6.0 runtime, as existing scripts can adopt the new version with minimal code changes. Data engineering teams running complex semi-structured ingestion should migrate target schemas to Iceberg v3 tables to capitalize on automatic VARIANT shredding for faster downstream query performance and lower storage footprints. However, platform leads must test custom dependencies against Python 3.13 and Spark 4.1 before rolling out fleet-wide auto-upgrades across production pipelines.