Larry Ellison Identifies Major Issue Affecting AI Models like ChatGPT, Gemini, Grok
Oracle co-founder and CTO Larry Ellison has identified a significant issue affecting the current landscape of artificial intelligence (AI) models, including popular systems like ChatGPT, Gemini, and Grok. According to Ellison, the reliance on publicly available data has transformed AI into a commodity, risking diminished differentiation among competing technologies.
Commoditization of AI Models
During Oracle’s fiscal Q2 2026 earnings call in December, Ellison highlighted that major AI models from companies such as OpenAI, Anthropic, Meta, and Google are primarily trained on the same dataset sourced from the internet. He pointed out this common foundation results in a lack of uniqueness among the models, leading to the rapid commoditization of advanced AI solutions.
The Quest for Proprietary Data
Ellison believes the future of AI lies in utilizing proprietary enterprise data rather than solely public data. He stated that the next significant opportunity will focus on enabling AI systems to work securely with private data, a sector he estimates will be even more lucrative than the current growth in GPU markets and data centers.
- Private Data Advantage: Oracle plans to leverage its extensive database capabilities to maintain a competitive edge.
- Investment Surge: The company has increased its capital expenditure projection to approximately $50 billion for the year, up from $35 billion just three months prior.
- AI Data Platform: Oracle’s AI Data Platform employs techniques such as Retrieval-Augmented Generation to allow real-time queries of private data without compromising security.
Competitive Landscape
Oracle remains committed to its ambitious AI infrastructure projects, with significant developments announced at Oracle AI World in October. These include:
- A 50,000-GPU supercluster utilizing AMD MI450 chips, set to launch in Q3 2026.
- The OCI Zettascale10 supercomputer, expected to interconnect hundreds of thousands of NVIDIA GPUs.
- As of late 2025, Oracle’s cloud backlog reached over $500 billion, largely fueled by AI demand.
However, Ellison’s strategy contends with several challenges. The rise of synthetic data generation may decrease the dependence on unique proprietary datasets. Additionally, competitors like Amazon Web Services, Microsoft Azure, and Google Cloud are also escalating their efforts to develop enterprise AI solutions.
The key question remains: will Oracle’s established position in enterprise databases provide the necessary leverage, or will the competitive landscape evolve faster than the realization of Oracle’s ambitious infrastructure investments?
