Oracle debuts Iceberg-compatible Autonomous AI Lakehouse to boost enterprise analytics

Oracle has unveiled a new data lakehouse offering that supports Apache Iceberg to help enterprises streamline analytics and enhance data interoperability across cloud and on-premises environments.

Enterprises focused on leveraging data for AI and analytics use cases have lately been turning to data lakehouses as they blend the affordability and scalability of data lakes with the performance of data warehouses.

The new offering, which has been named Autonomous AI Lakehouse and announced at Oracle’s ongoing AI World conference, combines the company’s existing services, such as Autonomous Database, Autonomous Data Warehouse, and Select AI, with new features, such as Apache Iceberg support, a query accelerator, a new database catalog — the Autonomous AI Data Catalog, and plug and play SQL support.

Focus on flexibility and interoperability

The support for Iceberg, as well as a database catalog that unifies data discovery and simplifies metadata management across clouds, data lakes, databases, shares, and catalogs, marks a change in Oracle’s strategy: from a single cloud model to recognizing enterprises’ need for flexibility across clouds and platforms, according to Moor Strategy and Insights principal analyst Robert Kramer.

“Oracle has shifted to open, vendor-neutral interoperability, aligning more closely with Microsoft Fabric and Google BigLake. Features such as Iceberg support and plug-and-play SQL integration with catalogs like Databricks Unity, AWS Glue, and Snowflake Polaris make it easy to support multi-cloud and hybrid setups across OCI, AWS, Azure, Google Cloud, and Exadata Cloud@Customer,” Kramer said.

Analysts also pointed out that Oracle’s new focus on interoperability will benefit its existing customers, especially those who are looking to adopt “open-table” lakehouse analytics without abandoning Autonomous Database or Exadata.

“For existing customers, the enhancements inside Autonomous AI Lakehouse translate into reduced integration complexity, minimal manual tuning, and a significantly faster path from data insights to business action,” said Michael Ni, principal analyst at Constellation Research.

“These enhancements are also reflective of Oracle’s broader strategic vision: transforming every database into an AI engine and every AI engine into an autonomous decision-making platform,” Ni added.

The principal analyst further feels that the interoperability-focused enhancements inside Oracle’s new lakehouse offering is a testament to the company’s realization that it must match rivals like Google and Microsoft on core data stack capabilities to compete for the next wave of enterprise IT spending.

However, HyperFRAME Research’s practice leader of AI Stack, Stephanie Walter, pointed out that getting customers for the new lakehouse would require serious evaluation, as several similar offerings with interoperable features are already available.

New customers will evaluate Oracle’s Iceberg interoperability, catalog breadth, governance, and price-performance on Exadata before making any decision on adoption, Walter said.

“When it comes to developers adopting the lakehouse, they will look for the depth of SQL on external Iceberg and the maturity of agent tooling,” Walter added.

Focus on increasing efficiencies and reducing costs

In contrast to Walter, Moor Insights and Strategy’s Kramer said that new customers wanting to try out Oracle’s new lakehouse might find some of its features appealing as they focus on increasing efficiencies and reducing costs.

The Data Lake Accelerator, for example, Kramer said, dynamically scales compute and network resources based on query demand and only charges for the resources used, essentially reducing compute waste.

“Early access testers cited significant improvements in query times and workload efficiency, with the ability to scale resources on demand. This is similar in intent to Databricks’ serverless SQL or Snowflake’s auto-scaling, but Oracle applies it to queries against external Iceberg data, which helps maintain performance without moving data,” Kramer explained.

Similarly, Exadata Table Cache, another feature inside Autonomous AI Lakehouse, can also help reduce costs.

“The Exadata Table Cache can improve query response times by keeping frequently accessed Iceberg data in Exadata’s flash storage. This idea resembles caching techniques in Snowflake and Databricks but benefits from the predictable performance of Exadata’s storage tiers, especially for mixed on-prem and cloud workloads,” Kramer said.

Other features, such as GoldenGate for Iceberg, which adds real-time data streaming into Iceberg tables from both operational and analytic systems, can also improve efficiency by reducing batch processing overhead, Kramer added.

The Autonomous AI Lakehouse also integrates one agent framework and an AI agent — the Select AI Agent and the Data Science Agent.

The Select AI Agent, according to Oracle, will provide a framework for enterprises to build, deploy, and manage AI agents within the lakehouse’s database.   

“It supports custom and pre-built PL/SQL tools, external tools via REST, and MCP Servers, enabling the automation of multi-step agentic workflows,” Oracle said in a statement.

Examples of agents that can be built using the framework include data analysis agents, data preparation agents, or operational workflow agents that alert or update records, HyperFRAME Research’s Walter said.

The Data Science Agent, as the name suggests, is a pre-built agent to help data practitioners explore data and generate insights in natural language. Oracle has made the Autonomous AI Lakehouse generally available. The pricing of the platform will vary based on the number of services being consumed by an enterprise, the company said.