Hudi 129 remains an active Apache project (1.0 in December 2024, 1.2.0 in May 2026, 1.2.1 in September 2026), with features the others lack: built-in table services that compact, cluster and clean as part of the writer, non-blocking concurrency control for many streaming writers, partial updates, and incremental queries that return only records changed since a given instant. It is strongest in CDC-heavy pipelines on Spark 129 and Flink 129 , and Onehouse, founded by Hudi's creator, sells it as a managed service.
The market, though, has largely standardized on Iceberg 129 for multi-engine access, with Delta Lake 237,929 as the Databricks-centered alternative: Snowflake, AWS 24 S3 Tables, BigQuery 1 and DuckDB 61,228 lead with Iceberg, and Databricks 2,717 bought Tabular, the company of Iceberg's creators, in 2024. Hudi's answer is interoperability: Apache XTable 129 (https://github.com/apache/incubator-xtable 1,254 ) (incubating, 0.4.0 in August 2026), started by Onehouse, translates table metadata among Hudi, Delta and Iceberg without copying data, much as UniForm does for Delta (UniForm).