Microsoft's Nadella Advocates for "Sovereign AI"

Microsoft CEO Satya Nadella has argued that enterprises need a more balanced and less dependent relationship with providers of frontier AI models. Nadella is proposing a "sovereign AI" model, where organizations own and control their own data, models, and AI infrastructure.

According to Nadella, current practices require companies to share significant amounts of proprietary knowledge and institutional learning with third-party models to benefit from them. He describes this as "paying for intelligence twice: once with money, and again with something even more valuable: the proprietary knowledge you must reveal to make that intelligence useful." He added, "The better you want the model to perform, the more of that knowledge you have to feed it!"

Nadella's proposal suggests that enterprises' data, generated outputs, evaluations, adapted model weights, and memory would accumulate and improve within the organization. The goal is to ensure that the company's unique knowledge and learning remain its own property, rather than benefiting competitors who might leverage the same information through third-party models.

This "sovereign AI" model appears to conflict with business models like OpenAI's, which offer access to their models as a managed service. This is particularly noteworthy given Microsoft's substantial investment in OpenAI. Nadella suggests that enterprises will increasingly demand the right to use the outputs of third-party models to train their own private models through a process called "distillation," a practice many providers currently forbid.