Unpacking Oracle’s AI database update – News

Oracle Corp. is architecting an AI database to transform enterprise workloads, simplify development and safeguard data.

The company is doubling down on its belief that “AI changes everything.” Its recent Oracle AI Database 26ai update marks a pivotal shift in how AI is integrated into the data stack. The release builds on decades of database innovation while infusing AI across every layer — from transactions and queries to application development and data governance, according to Juan Loaiza (pictured, right), executive vice president of database technology at Oracle.

“We launched a new major version of our database,” he said. “It’s almost there. If you’re already on our 23AI version, you just get it. When you apply your quarterly patch, you’re just going to become 26ai. It’s a super easy migration because we’re basically just layering AI on top.”

Loaiza spoke with theCUBE’s Dave Vellante (left) during an exclusive conversation with theCUBE, News Media’s livestreaming studio. They discussed Oracle’s decision to incorporate AI heavily in its database following the recent version update. (* Disclosure below.)

The AI database in detail

“AI Database” is not a marketing slogan, but a foundational redesign, according to Loaiza. Oracle has embedded AI vectors, large language model integration and automated retrieval-augmented generation directly into the database. The result: Users can perform semantic searches, find similar objects and run AI-enhanced queries without external tools or separate vector databases.

“We put a lot of new technologies for AI in there, and then moving on all the way into Lakehouse … and running AI queries against the existing database,” Loaiza said. “We have over a hundred projects in AI in this release of the Oracle Database. It’s a fresh new database that’s been completely infused with AI, not just in one place, but throughout the entire stack. Everything about the database we’ve infused AI with.”

Oracle’s AI Database also unifies AI capabilities across diverse data types — relational, spatial, document, graph, IoT, blockchain and streaming. This unified model eliminates the complexity of managing multiple AI tools for different workloads. With AI built into the core, users can add a single line of SQL to enable AI-powered search and analysis across all data forms.

“It makes the world dramatically simpler if you just have one AI technology, one AI architecture. It’s architected into the data — you don’t really have to do a whole lot,” Loaiza said. “That’s the other thing that’s interesting. Because we architected AI into the database, it’s super simple to use and easy to adopt.”

Oracle is addressing one of AI’s biggest enterprise hurdles — data privacy. As generative AI begins writing application code, ensuring strict access controls becomes critical. Oracle’s approach moves privacy enforcement into the database layer itself. This architecture prevents data leakage by design, ensuring compliance for industries handling sensitive financial or medical records, according to Loaiza.

“The AI only sees the data that the end user is allowed to see,” he said. “It’s the same thing with a lot of other business rules like data integrity, data locking, data validation and data evolvability. A lot of these things we’ve built into the app. With the AI generating it, and you must guarantee it, it’s much better to move those capabilities into the database where we can guarantee it underneath the AI.”

Here’s theCUBE’s complete video interview with Juan Loaiza:

(* Disclosure: Oracle Corp. sponsored this segment of theCUBE. Neither Oracle nor other sponsors have editorial control over content on theCUBE or News.)

Photo: News