Oracle’s Larry Ellison believes the AI race is becoming a commodity due to shared internet data. He argues the future lies in leveraging private enterprise data, a move Oracle is heavily investing in with massive infrastructure plans. While competitors are also building enterprise AI, Oracle’s existing database dominance could be key to its success in this next AI wave.

Oracle cofounder and CTO Larry Ellison has put his finger on what he sees as the fatal flaw in today’s AI race: every major model, from ChatGPT to Gemini to Meta’s Llama, is trained on essentially the same publicly available internet data. During Oracle’s fiscal Q2 2026 earnings call in December, Ellison argued this shared foundation is rapidly turning cutting-edge AI into a commodity product with razor-thin differentiation.“All the large language models—OpenAI, Anthropic, Meta, Google, xAI—they’re all trained on the same data. It’s all public data from the internet,” Ellison said. “So they’re all basically the same. And that’s why they’re becoming commoditized so quickly.”

The real money is in private data, not public models

Ellison’s solution? The next gold rush won’t be building better foundational models—it’ll be enabling AI to work with proprietary enterprise data while keeping it secure. He estimates this second wave of AI will prove “even larger and more valuable” than the current boom in GPUs and data centers.Oracle is betting heavily on this vision, projecting roughly $50 billion in capital expenditures for the full year, up from $35 billion estimated just three months earlier. The company argues it has a natural advantage since most high-value private data already lives in Oracle databases. Its AI Data Platform uses techniques like Retrieval-Augmented Generation to let models query private information in real time without security compromises.

Oracle’s massive AI infrastructure bet faces stiff competition

The company announced partnerships at Oracle AI World in October, including a 50,000-GPU supercluster with AMD MI450 chips launching in Q3 2026 and the OCI Zettascale10 supercomputer connecting hundreds of thousands of NVIDIA GPUs. By late 2025, Oracle’s cloud backlog had topped $500 billion, driven mostly by AI demand.Still, Ellison’s thesis faces headwinds. Synthetic data generation could reduce reliance on exclusive proprietary datasets, while rivals like Amazon Web Services, Microsoft Azure, and Google Cloud are racing to build similar enterprise AI capabilities. The question is whether Oracle’s existing grip on enterprise databases will prove decisive—or if the AI landscape will shift before the company’s massive infrastructure bets pay off.