Oracle founder and CTO Larry Ellison has identified what he calls the critical limitation holding back AI models from Gemini to ChatGPT: they’re all trained on the same publicly available internet data, making them increasingly commoditized with little real differentiation between them.Speaking during Oracle’s fiscal Q2 2026 earnings call in December, Ellison argued that while training foundational models on public data has created “the largest and fastest-growing business in human history,” the real value lies ahead in a second phase: enabling these models to securely reason over private enterprise data without compromising security.“For these models to reach their peak value, you need to train them not just on publicly available data, but make privately owned data available for those models as well,” Ellison said. He estimates this next phase will prove “even larger and more valuable” than the current GPU and data center boom.
Oracle bets big on becoming the enterprise AI backbone
Oracle is positioning itself as uniquely equipped for this shift. The company points out that most of the world’s high-value private data already lives in Oracle databases. Its AI Data Platform uses techniques like Retrieval-Augmented Generation to let any major AI model query private data in real-time while maintaining security.The company is backing this vision with aggressive spending, projecting roughly $50 billion in capital expenditures for the full year, up from $35 billion estimated in September. At Oracle AI World in October, the company announced partnerships including a 50,000-GPU AI supercluster with AMD MI450 chips launching in Q3 2026, and the OCI Zettascale10 supercomputer connecting hundreds of thousands of NVIDIA GPUs.
The race for private data access heats up
Ellison’s thesis faces challenges from multiple angles. Synthetic data generation could reduce dependence on exclusive proprietary data sets; real-time user interaction data from consumer apps could turn out to be more valuable than static enterprise records. At the same time, rivals like Amazon Web Services, Microsoft Azure and Google Cloud are racing to develop comparable enterprise AI capabilities, though Oracle argues its established grip on enterprise databases gives it a strategic advantage. By late 2025, the company’s cloud backlog had topped $500 billion, driven mostly by demand for AI.
About the Author
