Google Bolsters Apache Iceberg Interoperability, Unveiling Cross-Cloud Lakehouse Capabilities at Recent Summits

Google has significantly advanced its commitment to open data lakehouse architectures, announcing a suite of new interoperability features for Apache Iceberg within its BigQuery platform. These developments, unveiled at the Apache Iceberg Summit and further expanded at the recent Google Next ’26 conference, aim to dissolve data silos, reduce operational complexities, and enhance the utility of data for advanced analytics and artificial intelligence workloads across hybrid and multi-cloud environments. The cornerstone of these announcements is the preview of a serverless Iceberg REST catalog, designed to empower data teams to create, update, and query the same Apache Iceberg tables seamlessly across BigQuery and other popular compute engines like Spark, Flink, and Trino, all without the need for data duplication.

The Foundation: Enhanced Apache Iceberg Support in BigQuery

The initial set of announcements at the Apache Iceberg Summit centered on making BigQuery a more versatile hub for Iceberg-based lakehouses. At its core, the preview introduces a serverless Iceberg REST catalog, a critical component for achieving true interoperability. This catalog acts as a central metadata store, allowing different query engines to understand and interact with the same underlying Iceberg tables. Previously, organizations leveraging Apache Iceberg for their data lakehouse architectures often faced a dilemma: either manage their Iceberg tables through a Google-managed Iceberg REST catalog, or opt for tables managed directly by BigQuery. This created a fragmentation where, for instance, customers relying on Apache Spark for Extract, Transform, Load (ETL) operations into Iceberg REST Catalog tables couldn’t fully utilize BigQuery’s native write capabilities or its robust storage management features. This new unified approach eliminates that choice, providing a more cohesive experience.

Beyond basic access, Google is introducing managed support for several critical operational aspects of Iceberg deployments. This includes automated metadata management, which is crucial for maintaining the integrity and discoverability of data across diverse tools. Furthermore, BigQuery will now offer managed services for table maintenance tasks, such as compaction and garbage collection, which are often manually intensive and error-prone processes in self-managed Iceberg environments. These features are designed to offload the heavy lifting from data platform teams, allowing them to focus on data utilization rather than infrastructure upkeep. The overall goal, as articulated by Yuriy Zhovtobryukh, Senior Product Manager at Google, and Angela Soares, Senior Product Marketing Manager at Google, is to simplify the lakehouse journey: "If you’re building a lakehouse today, you’re probably using Apache Iceberg, which has gained massive popularity among data platform teams that need to support multiple compute engines (like Spark and BigQuery) that access the same data for different workloads." Their statement underscores the growing demand for Iceberg’s capabilities and Google’s strategic response to it.

Google Next ’26: Expanding to a Cross-Cloud, AI-Ready Lakehouse

Building on the initial announcements, Google significantly broadened the scope of its Iceberg interoperability at the Next ’26 conference, unveiling a vision for a truly cross-cloud lakehouse that seamlessly integrates with advanced AI workflows. This expansion represents a strategic move to position Google Cloud as a central orchestrator for data residing across various cloud providers and external data platforms. The key highlight here is the support for querying Iceberg catalogs across major cloud environments, including Amazon Web Services (AWS) and Microsoft Azure, as well as popular data platforms like Databricks and Snowflake. This multi-cloud capability directly addresses the prevalent enterprise reality where data often resides in distributed environments, avoiding vendor lock-in and promoting data fluidity.

Google’s overarching objective with these expanded capabilities is to empower organizations to maintain their data in open formats, thereby retaining maximum flexibility, while simultaneously leveraging a diverse array of processing and analytics tools on the same datasets. This commitment to open standards is a core tenet of the modern data ecosystem, allowing enterprises to choose the best tool for each specific workload without being forced into proprietary data formats or vendor-specific ecosystems. The integration with AI workflows is particularly salient, reflecting the industry’s accelerating shift towards data-driven AI. By enabling direct access to Iceberg tables from AI tools and frameworks, Google is paving the way for more efficient and robust machine learning model training and deployment.

Addressing the "Hidden Tax" of Apache Iceberg Adoption

Despite Apache Iceberg’s compelling technical merits and growing popularity, many teams adopting it still encounter significant challenges that translate into higher costs and operational complexities. These challenges are particularly pronounced in areas like streaming data ingestion, building reliable replication pipelines, and establishing consistent governance across a multitude of tools. Google argues that compared to fully managed data platforms, the self-managed Iceberg experience can be arduous.

To mitigate these issues, Google is extending its robust BigQuery infrastructure to natively support Iceberg tables. This means that Iceberg users can now benefit from BigQuery’s battle-tested capabilities, including:

• Managed Metadata: Automated handling of schema evolution, partitioning, and table versioning, reducing manual oversight.
• Automatic Table Maintenance: BigQuery will automatically perform tasks like data compaction and garbage collection, optimizing query performance and storage efficiency without requiring human intervention.
• Transactions: Ensuring ACID (Atomicity, Consistency, Isolation, Durability) properties for data modifications, critical for data integrity, especially in concurrent write environments.
• Change Data Replication: Streamlining the process of capturing and applying changes to Iceberg tables, crucial for real-time analytics and data synchronization.

This integrated approach directly addresses the "hidden tax" on Iceberg adoption, a term often used by practitioners to describe the unforeseen operational overhead. David Colbert, a recognized voice in the data community, aptly summarized this friction: "Teams get excited about Iceberg/Delta capabilities but hit friction fast on compaction, metadata management, and orchestration. The catalog point is key. Open formats solve storage portability, but control plane choices determine long-term optionality." Google’s managed services aim to remove these common friction points, allowing data teams to realize Iceberg’s benefits without getting bogged down in infrastructure management. The centralized table access controls included in the preview further enhance governance, allowing permissions to be managed consistently across various query engines, a critical feature for data security and compliance in complex enterprise environments.

Enabling Modern Data and AI Workflows with Integrated Tools

The expansion of Iceberg interoperability is not just about making tables accessible; it’s about making them useful for the most demanding modern workloads, especially those involving artificial intelligence. Google has introduced several key features to facilitate this:

• BigQuery ObjectRefs (Generally Available): This feature allows teams to seamlessly combine structured Iceberg data with unstructured files stored in Google Cloud Storage. This capability is pivotal for multimodal analysis, where insights are derived from diverse data types (e.g., combining customer transaction data from Iceberg with product images or customer service call recordings from Cloud Storage). Such integration is a prerequisite for many advanced AI and machine learning applications that require a holistic view of data.
• Knowledge Catalog (formerly Dataplex, in Preview): Positioned as a comprehensive governance layer, Knowledge Catalog is designed to manage metadata, data lineage, and access controls across disparate systems, including the newly integrated Iceberg tables. This centralized governance is essential for maintaining data quality, ensuring regulatory compliance, and providing data discoverability within large organizations. It acts as a single pane of glass for understanding and controlling data assets across the entire data estate.

These tools, combined with the cross-cloud Iceberg capabilities, align with Google’s broader strategy for the "agentic era" – an era where AI agents autonomously reason over vast datasets. Precious Pendo, commenting on the Next ’26 announcements, insightfuly noted: "Google is betting that enterprise AI value will accrue to whoever owns the reasoning layer over data, not just the storage layer. AWS and Azure charge you for compute and storage. Google wants to charge you for context and intelligence." This perspective highlights Google’s ambition to move beyond commodity cloud services and establish itself as a leader in providing the intelligent infrastructure required for next-generation AI.

The Broader Lakehouse Ecosystem and Competitive Landscape

Apache Iceberg’s journey from a Netflix engineering project to an undisputed standard for open data lakehouse architecture in less than seven years is a testament to its technical superiority and the industry’s need for robust, open table formats. As Shashank Muthuraj, a cloud engineer at Red Oak Strategic, aptly puts it: "The technical merits – ACID transactions, hidden partitioning, time travel, and engine independence – are compelling, but the real story is the unprecedented industry alignment." Iceberg’s ability to provide ACID guarantees, manage schema evolution, enable time travel for historical analysis, and offer performance optimizations like hidden partitioning, all while remaining engine-agnostic, has made it a cornerstone of modern data architectures.

Google Cloud is certainly not alone in recognizing Iceberg’s importance. Major cloud providers and data platform vendors are actively integrating and supporting Iceberg workloads. AWS, for instance, offers native support for Iceberg across several of its analytics services, including Amazon EMR, AWS Glue, Amazon Athena, and Amazon Redshift. This competitive landscape underscores the strategic significance of Iceberg in the evolving data ecosystem. Google’s approach, however, emphasizes not just native support but also deep interoperability across clouds and external platforms, aiming to provide a flexible and comprehensive solution that transcends individual vendor boundaries. By enabling querying across AWS and Azure, and interoperability with platforms like Databricks and Snowflake, Google is positioning BigQuery as a powerful, multi-cloud data orchestration layer, allowing customers to unify their data strategy even if their data remains distributed.

Current Availability and Future Outlook

While the core managed Apache Iceberg table support within BigQuery is now generally available, signifying its readiness for production workloads, the broader open interoperability features and the Iceberg REST catalog capabilities announced at the Iceberg Summit and Google Next ’26 are currently in preview. This phased rollout allows Google to gather feedback from early adopters and refine the services before general availability. The ongoing investment in Iceberg and the commitment to open standards signals Google’s long-term vision for a flexible, scalable, and AI-ready data infrastructure. As data volumes continue to explode and the demand for real-time insights and advanced AI applications grows, open lakehouse architectures, powered by formats like Apache Iceberg and integrated seamlessly into managed cloud services, will be critical enablers for enterprise innovation. Google’s latest announcements position BigQuery as a formidable player in this evolving landscape, offering a compelling blend of openness, managed services, and multi-cloud reach.