Nadella: Every Company Should Build Its Own AI Model

Microsoft CEO Satya Nadella has articulated a bold vision for enterprise AI: every company should develop its own artificial intelligence models rather than depending exclusively on a small number of foundation model providers.

"My simple thing is there should be as many models in the world as firms in the world," Nadella said in an interview with Yash Patil, cofounder of Applied Compute, published Friday. "Because after all, what is a firm? A firm is a learning system."

Why it matters

Nadella's comments challenge the current trajectory of enterprise AI adoption, where most organizations consume models from OpenAI, Anthropic, Google, and Meta. His argument frames AI development not as a technical procurement decision but as fundamental to organizational identity and competitive differentiation. If companies outsource their core learning capabilities, Nadella suggests, they risk losing the very characteristics that justify their existence.

The case for custom models

Nadella emphasized that organizations need flexibility to use their proprietary data and operational context. "I don't want to be locked into any one model," he explained. "I want to be able to use my own context, my own data — in fact, my own traces to maybe even take a much more open-weight, cost-efficient model or a fine-tuned model."

The Microsoft chief warned that concentration in AI capabilities creates systemic economic vulnerabilities. "It can't be, 'Hey, look, I have two frontier models or three frontier models' or whatever, some finite set that have learned everything that is differentiated today in the economy because then it collapses," Nadella said.

He drew a distinction between purchasing tools and outsourcing fundamental capabilities. "You can always buy a tool, you can even outsource a task or even a job, but you can't outsource your learning," Nadella said. "If you outsource your learning, then why exist?"

Microsoft's multi-model approach

Microsoft has backed this philosophy through Azure AI Foundry, which hosts models from multiple providers including OpenAI, DeepSeek, and Cohere rather than relying exclusively on its high-profile OpenAI partnership. This mirrors strategies from Amazon with Bedrock and Google Cloud, which offer catalogs of third-party and proprietary models alongside their flagship offerings.

Many enterprises are already experimenting with open-weight models—AI systems with publicly available parameters that allow companies to fine-tune and deploy them independently. Examples include Meta's Llama and Mistral's model families.

The comments represent one of Nadella's clearest articulations yet of how he believes enterprise AI should evolve beyond the current reliance on a handful of foundation model providers.

These details were first reported by Business Insider.