[Digital Today reporter Jinju Hong] Microsoft CEO Satya Nadella (사티아 나델라) warned that if economic value is concentrated in a small number of AI companies, industries across the economy could lose control over knowledge and be weakened.

Business Insider reported on June 15 that Nadella wrote in a post on X on June 14 that if the structure in which corporate knowledge is absorbed into AI models becomes entrenched, only some model providers will benefit and each industry could lose its accumulated learning assets.

Nadella said the scenario he worries about most is a structure in which companies across all industries hand over value to a small number of models. "None of us want a world where every company in every domain gives up value to a few models that eat everything in sight," he wrote. He added, "There is no social license for an AI future that hollows out entire industries."

The remarks show that the AI race is shifting beyond model performance to the question of who controls data and corporate work knowledge. Nadella presented a broad AI ecosystem in which companies continue to control their own learning systems as an alternative. He argued that innovation can continue only if companies maintain internal learning systems and employee expertise.

Nadella compared the risk to the early days of globalization. "Think about how in the first stage of globalization, outsourcing hollowed out entire industrial economies," he wrote, adding that while gross domestic product figures may have looked fine, replacement and departures happened and the consequences continue today. He pointed out that even if AI adoption appears as growth in macro indicators, knowledge and roles can drain away on the industrial front lines.

A similar view has been voiced by other big tech executives. Sridhar Ramaswamy (스리다르 라마스와미), CEO of Snowflake, said on a podcast in February that large software companies could be pushed back into being simple data suppliers. "The big model developers want to create a world where all the data from every company is easily provided to them," he said, warning that "the rest of the world could end up as a dumb data pipe feeding data to a huge brain." He also said he was concerned users may come to want an integrated agent that contains all data from Snowflake and elsewhere, instead of AI agents made by software companies.

Box CEO Aaron Levie (에런 레비) also wrote in a LinkedIn post in January that AI models can perform high-level knowledge work in almost all professions, including law, strategy and scientific research. He said the key challenge will be how companies differentiate themselves in an environment where everyone has access to the same level of expert intelligence. Levie suggested context as the answer.

The trend shows that the axis of competition in the AI market is shifting from the models themselves to company-held data, work processes and in-house expertise. Nadella's remarks also focus on the idea that even if companies adopt AI, they should not hand over core knowledge and learning structures entirely to external models. Who builds models, who controls data and how much companies maintain their own learning systems are expected to emerge as key issues in the AI ecosystem going forward.