Trending Topics
Trending
Oracle’s Larry Ellison outlines AI future built on secure access to private enterprise data
“AI changes everything.” These were almost the first words uttered by Larry Ellison, Oracle’s Co-Founder and CTO, during his keynote address at Oracle’s AI World 2025 conference. And he didn’t waste any time making the case for leveraging enterprise and private data to train and deploy large multimodal AI models to surpass its competitors in the race to solve very real human problems.
Central to Ellison’s forward-looking vision for Oracle is the company’s big step into AI database architecture, anchored by a redesigned Oracle Database and a new AI Data Platform. One that uses retrieval-augmented generation (RAG) and vector search to let large language models securely reason across both public and private data.
Ellison positioned the Oracle AI as a re-engineered evolution of the company’s flagship system that enables enterprises to access and apply information stored across Oracle and other clouds without exposing it.
In practical terms, organisations can now connect their proprietary data to any major AI model while keeping ownership and compliance intact. The result is a platform built for reasoning rather than storage, turning enterprise data into immediate, actionable intelligence, Ellison explained.
Oracle’s new 1.2-billion-watt AI brain
To support Oracle’s multimodal AI model training ambitions, Ellison showcased the company’s massive expansion of its physical footprint: a sprawling data centre campus in Abilene, Texas. Described as “a 1.2-billion-watt AI brain”, the 1,000-acre compound hosts more than 450,000 more than 450,000 Nvidia GB200 Grace Hopper Superchips powered by the local grid and onsite natural gas turbines.
“We trained the first version of Grok,” said Ellison. “We are certainly involved in training more multimodal AI models than any other company. This new campus will be where we do a lot of that training – for us and for our partners.”
“Our AI models are writing”
Shifting from scale to application, Ellison shared that Oracle’s new AI capabilities are being put to use inside the company, with its AI models now generating production code, building customer insights, and automating outreach.
Oracle has also begun using RAG across its own customer data to predict buying behaviour, select references, and compose personalised sales emails.
“A lot of the code Oracle is writing – Oracle isn’t writing,” Ellison said. “Our AI models are writing.”
Healthcare as the new AI frontier
While enterprise automation evolution was at the core of Ellison’s keynote, he framed revolutionising healthcare as the next frontier for Oracle’s AI systems. The company is rebuilding its Cerner platform using AI while integrating hospitals, insurers, regulators, and banks into a connected ecosystem.
“The AI agent reasons with all of this data to propose the best care at the highest reimbursement level achievable,” he said.
Ellison added that the system will even allow banks to lend against healthcare receivables, extending AI’s reach from patient care to finance.
After 90 minutes outlining Oracle’s AI future, Ellison’s message was clear: the company is betting that control of enterprise information will define the next phase of the AI race.
Related articles
