Oracle founder Larry Ellison explains the biggest problem with AI models from Google, OpenAI, Meta and other big tech companies – The Times of India

Oracle founder Larry Ellison explains the biggest problem with AI models from Google, OpenAI, Meta and other big tech companies – The Times of India

By TOI Tech Desk
Publication Date: 2026-01-09 08:14:00

Oracle’s Larry Ellison believes the next AI challenge is private corporate data, not just the public Internet. He argues that current models are becoming commodified and the real value lies in securely accessing and reflecting on sensitive corporate information. Oracle is investing heavily and leveraging its extensive database presence to become the company’s AI backbone and faces competition from cloud rivals.

Larry Ellison, founder and CTO of Oracle, has identified what he believes is the key limitation holding back AI models from Gemini to ChatGPT: They are all trained on the same publicly available internet data, making them increasingly commodified and with little difference between them.During Oracle’s second quarter 2026 earnings call in December, Ellison argued that while training fundamental models on public data has created “the largest and fastest-growing company in the history of humanity,” the real value lies in a second phase: enabling those models to reason securely on private corporate data…