Iceberg Summit 2026: The Open Table Format That's Powering the Next Generation of Data Lakehouses

Last week (April 8-9, 2026), over a thousand data engineers, architects, and AI builders gathered at the San Francisco Marriott Marquis for Iceberg Summit 2026-the premier in-person event dedicated to Apache Iceberg. Now in its third edition and officially sanctioned by the Apache Software Foundation, the summit delivered two packed days of workshops, technical deep-dives, keynotes, and hallway conversations that made one thing crystal clear: Apache Iceberg is no longer just an "open table format." It has become the de facto foundation for modern, open data lakehouses at scale.

From keynotes by Russell Spitzer (Apache Iceberg PMC, Snowflake) and Julien Le Dem (Apache Parquet PMC Chair, Datadog) to sessions on massive-scale table layouts, AI-ready governance, multi-cloud interoperability, and real-world production stories from companies like Apple, Bloomberg, Pinterest, and Wells Fargo, the energy was electric. The message echoed throughout the venue: open formats are moving from "nice-to-have" to mission-critical infrastructure for AI at scale.

Why Iceberg Summit 2026 Felt Like a Turning Point

The data world has moved on from "lake vs. warehouse" debates. Teams now demand flexibility and reliability: real-time streaming, ACID transactions, schema evolution, time travel, petabyte-scale performance, and zero vendor lock-in. Iceberg delivers it all-and the summit showed the community pushing the boundaries even harder. Standout moments included:

Day 1 Keynote by Russell Spitzer (Apache Iceberg PMC & Principal Engineer at Snowflake): "From Batch to Streaming and AI, Iceberg for Everyone by Everyone" - a masterclass on how Iceberg is erasing the lines between batch, streaming, and AI workloads. (Watch the spotlight teaser here)

Day 2 Keynote by Julien Le Dem (Apache Parquet PMC Chair & Principal Engineer at Datadog): A forward-looking dive into how AI is reshaping columnar data ecosystems and what it means for Iceberg's future. (Spotlight here)

Real-world war stories poured in from Apple, Bloomberg, Pinterest, Wells Fargo, and more-covering massive-scale table layouts, hidden partitioning, AI-ready governance, multi-cloud interoperability, and the streaming + lakehouse convergence that everyone's chasing. The hallway buzz? "Iceberg is eating the data lakehouse world." (Catch an excellent community recap video summarizing the biggest trends here)

Session recordings and slides from the full 70+ talks are rolling out on the official Iceberg Summit YouTube playlist in the coming weeks-bookmark it: Iceberg Summit 2026 Playlist.

How Dataverses Is Building the Future on Apache Iceberg

We built our entire platform as a managed streaming data lakehouse with Apache Iceberg as the open foundation. Here's how we put Iceberg to work:

• Unified streaming + lakehouse storage: We combine Apache Kafka for real-time ingestion, Spark and Flink for processing, and Iceberg tables for durable, queryable storage. The result? A single platform where raw streaming events flow directly into ACID-compliant Iceberg tables-handling late-arriving data, exactly-once semantics, and sub-100ms query latency without any custom ETL pipelines.
• Declarative pipelines that just work: Our low-code YAML-based pipelines automatically create, optimize, and maintain Iceberg tables. Users define sources (50+ connectors including PostgreSQL CDC, MongoDB, Kafka, S3, etc.), transformations, and sinks once-and Dataverses handles schema evolution, compaction, snapshot management, and partitioning behind the scenes.
• Optimized for S3 and multi-cloud: We've implemented production-grade Iceberg optimizations on object storage (S3, GCS, Azure Blob) so customers get warehouse-like performance without the cost or complexity of proprietary data warehouses.
• AI AgentFlow Enterprise on live lakehouse data: Our drag-and-drop AgentFlow canvas lets teams build autonomous AI agents that query and act on fresh Iceberg data in real time. No more data silos or stale caches-agents operate directly on the governed lakehouse with full lineage and auditability.

We recently published deep dives on exactly these patterns: CDC and Slowly Changing Dimensions on Iceberg, Spark's real-time mode, and S3-specific optimizations. These aren't theoretical-they power production workloads processing billions of events daily for our customers.

What We're Taking Back to the Dataverses Roadmap

Iceberg Summit 2026 reinforced three priorities we're accelerating:

• Deeper Iceberg v4 integration (especially around new metadata and delete-file improvements).
• Even tighter streaming convergence so batch and real-time become truly indistinguishable.
• Enterprise governance features that make Iceberg the default choice for regulated industries.

The open data lakehouse isn't a future trend-it's here today. And the community driving it forward at events like Iceberg Summit is making it better every month.

Ready to Build Your Own Open Data Lakehouse?

If you're tired of wrestling with fragmented data stacks, expensive proprietary warehouses, or brittle streaming pipelines, we invite you to see what a true managed Iceberg-powered lakehouse feels like.

• Visit dataverses.io and book a demo
• Try our low-code streaming lakehouse in minutes-no infrastructure to manage
• Join the conversation on LinkedIn: we're sharing more Iceberg Summit takeaways and real customer stories every week

The future of data is open. At Dataverses, we're making that future ridiculously easy to use.

See you at the next Iceberg Summit-and in the lakehouse