Artificial intelligence is growing fast, and big tech companies like Google, OpenAI, Meta, and others are spending billions to build smarter AI models. But Oracle founder Larry Ellison believes there is a serious problem with today’s AI models. According to him, most of them are not very different from each other.
Larry Ellison said this during the Oracle’s fiscal Q2, 2026 earnings call in December. “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.
Same Training Data
Larry Ellison says the main issue is that almost all popular AI models are trained on the same public internet data. This includes websites, articles, forums, and other online content that anyone can access. Because the data is similar, the AI systems also behave in similar ways.
This is why many chatbots give almost the same answers, even though they are made by different companies. Larry Ellison believes that AI is slowly becoming a common product, where every model looks impressive but offers little real difference.
Why This Is a Problem?
When AI models depend only on public data, they cannot provide deep or exclusive insights. They may sound smart, but they are limited to what is already available online. Larry Ellison feels this limits innovation and reduces the real value of AI for businesses.
In simple terms, if everyone studies from the same book, their answers will also sound the same. That is exactly what is happening with most AI tools today.
The Future of AI Lies in Private Data
Larry Ellison believes the future of AI is not in public internet data but in private company data. This includes business records, customer information, financial data, and internal systems. AI that understands such data can give much more useful and accurate results.
Oracle is focusing strongly on this idea. The company wants to help businesses use AI on their own data while keeping it secure and private. According to Ellison, this is where AI can actually solve real problems and make a big difference.
Oracle’s Strategy Going Ahead
Instead of competing only on chatbots, Oracle is building AI tools for enterprises. These tools are designed to work inside companies and help with decision-making, planning, and operations.
Larry Ellison believes that the next phase of AI growth will depend on who controls valuable data, not just who builds the biggest model. In the long run, companies that combine AI will lead.
