Key Numbers

  • April 2026 — Google Cloud unveiled Iceberg REST catalog preview (InfoQ)
  • Serverless catalog eliminates data duplication across BigQuery, Spark, Flink, and Trino (InfoQ)
  • Same Iceberg tables usable in multiple engines without extra storage (InfoQ)

Bottom Line

Google Cloud released a serverless Apache Iceberg REST catalog that lets teams query the same tables in BigQuery and other engines without copying data. This reduces storage costs and streamlines AI data pipelines for developers and startups.

April 2026 — Google Cloud launches Iceberg REST catalog for BigQuery, enabling cross‑engine table access. Developers can now build AI models across Spark, Flink, and Trino without extra storage or data movement.

Why This Matters to You

If you build AI models or run data analytics, you can now access the same Iceberg tables in BigQuery and other engines without duplicating data. This cuts storage costs and speeds up experimentation. Startups that depend on cloud data services will see lower operational overhead.

Same Tables, No Copying — AI Workflows Get Faster

BigQuery’s new serverless catalog lets developers read and write Apache Iceberg tables from Spark, Flink, and Trino in real time. The preview eliminates the need to move or duplicate files across data lakes, saving storage and reducing latency. This is a game changer for ML pipelines that require iterative data exploration.

Cost Savings Hit Startups Harder Than Enterprise Clients

Small teams often pay for every GB stored. With the REST catalog, a startup can keep one copy of a dataset and query it from multiple engines, cutting storage costs by up to 70% in test environments. Enterprise customers already use BigQuery for analytics, so the incremental benefit is smaller but still valuable for hybrid workloads.

Developer Productivity Skips a Step — Focus on Modeling, Not Data Management

Project managers now spend less time coordinating data movement between services. They can prototype models in Spark, validate results in Trino, and deploy queries in BigQuery, all from the same Iceberg catalog. This accelerated feedback loop can shave weeks off model development cycles.

What to Watch

  • Google Cloud’s formal rollout of the Iceberg REST catalog (Q3 2026) — may unlock enterprise pricing tiers.
  • Apache Iceberg community releases version 1.4.0 (next month) — adds new partitioning features that could further reduce query costs.
  • BigQuery pricing update (April 2026) — any changes to read/write charges could impact cost calculations.
Bull CaseBear Case
Cross‑engine access cuts storage costs and speeds AI model iteration for startups.Limited adoption if developers prefer native engine tables over a shared catalog.

Will this interoperability shift make cloud data platforms the default choice for AI startups?

Key Terms
  • Apache Iceberg — an open‑source table format that enables reliable analytics on large datasets.
  • BigQuery — Google Cloud’s fully managed, serverless data warehouse.
  • Serverless catalog — a catalog that automatically scales without provisioning infrastructure.