As the tech industry continues to pour billions into Generative AI research, prominent software pioneers are raising flags about the long-term profitability of raw AI models. Sridhar Vembu, the founder and CEO of Zoho, has publicly backed Oracle co-founder Larry Ellison’s view that artificial intelligence models are rapidly becoming commoditized.

Reposting a video clip of Ellison’s recent commentary, Vembu shared his agreement on social media, writing: “AI is rapidly commoditizing. The value shifts to what is built around it.” This simple yet profound assertion highlights a growing consensus among veteran enterprise software developers.

Larry Ellison’s Core Argument: The Internet Data Ceiling

In the clip shared by Vembu, Oracle’s Larry Ellison explained why standard frontier AI models are losing their unique competitive edge:

• Identical Training Sets: Most leading AI models are trained on the exact same publicly available internet data, meaning their core cognitive baselines are highly similar.
• The Data Shift: Ellison argues that the true, sustainable competitive advantage for companies in the AI era will not come from the models themselves, but from having exclusive access to private, proprietary enterprise datasets.

Vembu’s Warnings on the AI Investment Bubble

Sridhar Vembu has been one of the most vocal pragmatists regarding the current tech landscape. While he strongly advocates for AI’s actual utility in software development, he has warned that the current market hype resembles a classic financial bubble.

His perspective outlines a distinct corporate path:

• Focus on Contextual Workflows: Instead of building massive general models, Zoho is focusing on integrating light, fast, domain-specific models into contextual business workflows.
• The Value of Orchestration: The real value lies in the software ecosystem, user interfaces, security protocols, and operational integrations built around the AI, rather than the raw model itself.

As AI models become cheaper, open-source, and readily accessible, Vembu and Ellison’s insights suggest that the future tech leaders will be those who control proprietary data and orchestrate workflows, rather than those who simply build the largest compute clusters.