Oracle Repositions Lakehouse Strategy Around AI Agent Data Access
As artificial intelligence systems become primary consumers of enterprise data, Oracle is restructuring its data platform to prioritize real-time access and semantic understanding across fragmented data sources.
The emergence of a unified semantic layer is increasingly defining which lakehouse platforms serve traditional analytics versus those engineered for artificial intelligence workloads.
When AI agents operate alongside human analysts accessing corporate information, organizations require infrastructure capable of interpreting data distributed across multiple cloud environments, storage formats and disparate systems. Over the past year, Oracle Corp. has overhauled its data infrastructure in response to this shift, as part of a larger initiative to modernize its database offerings. The company now presents its Autonomous AI Lakehouse as an interoperable layer that extends across an organization's existing data infrastructure without displacing it, according to Nipun Agarwal, senior vice president of software engineering for Oracle Database.
Lakehouses were the platform which customers use for running analytics. Now, lakehouses have become the decision backbone for AI.
Nipun Agarwal, Oracle
Agarwal shared these insights during an appearance at Oracle AI World, where he spoke with theCUBE's Dave Vellante in an exclusive broadcast. The conversation centered on how Oracle's Autonomous AI Lakehouse integrates a shared semantic layer with rapid, real-time data access capabilities across multiple platforms to enable AI agent functionality.
Building an open semantic layer across catalogs
Oracle's strategy depends on a catalog system capable of mounting external catalogs, including Databricks Inc.'s Unity Catalog, as well as Postgres and MySQL databases, allowing queries to execute against data in its original location rather than requiring duplication. At its October event, the company unveiled a business semantic layer alongside expanded Apache Iceberg compatibility, with automatically generated table and column metadata applying across non-Oracle data sources as well, according to Agarwal. Notably, Oracle positions this offering as complementary rather than competitive with alternative lakehouse platforms.
We really believe that for any enterprise, they're gonna have a choice. They're gonna be using multiple data sources. They'll have multiple lakehouses.
Nipun Agarwal, Oracle
Oracle's Lake Cache enhances query performance by caching the most frequently accessed Iceberg tables within Exadata's flash storage. Real-time data availability is becoming increasingly critical as software agents emerge as the dominant data consumers: Periodic refreshes of replicated datasets prove insufficient when agents need information on-demand, according to Agarwal.
https://www.youtube.com/embed/p5rbHfF_Rz4?feature=oembed
With AI that's no longer the case, because for AI, the consumer is not a human necessarily. The need for real-time access is a lot more with AI.
Nipun Agarwal, Oracle