Microsoft CEO warns: AI is replicating the tragedy of globalization, and every company must accumulate "human capital" + "Token capital"

Microsoft CEO Nadella warns that if AI value is concentrated in a few models, it will repeat the tragedy of global industry hollowing out, and proposes an ecological framework combining "human capital" and "token capital."
(Background: Amazon invests billions to build an AI data center in Missouri! Creating over 400 jobs, emphasizing "green water cooling" for sustainable operation)
(Additional context: The female stock goddess IPOs with $500 million to buy SpaceX on the first day! ARK Invest sells stocks to raise funds but still wants to buy)

Microsoft CEO wrote, "What we least want is a world where every company hands over its value to a few all-consuming models." Last night, Nadella (https://t.co/vLmiBKTtX3) posted a lengthy article on X titled "A frontier without an ecosystem is not stable," attempting to draw a moral red line for the development path of the AI industry.

https://t.co/vLmiBKTtX3

— Satya Nadella (@satyanadella) June 14, 2026

Two types of capital, one gamble

Nadella’s argument is built on a contrast between two concepts.

The first is human capital: the judgment, customer relationships, domain knowledge, and pattern recognition skills accumulated by employees over many years.

The second is token capital: the AI capabilities that a company trains, fine-tunes, or adjusts itself, without relying on external model providers. Simply put, it’s not just "renting AI to use," but integrating the company's own knowledge to build its own AI moat.

Nadella wrote on X that these two are not mutually exclusive. Human capital does not depreciate because of AI capabilities; instead, it becomes more valuable. He offers a key test to judge whether a company's AI deployment is successful: "Can it replace a general-purpose model without losing the expertise of its veteran employees?"

The essence of this question is whether a company has established its own "learning loop," continuously feeding its data and workflows back into the AI system, creating an increasingly intelligent knowledge moat, rather than repeatedly renting capabilities from external providers.

Nadella concludes: "Our priority must be to build a frontier ecosystem, not just a frontier model, so that value can flow to every company, every industry, and every country."

The ghost of globalization

Nadella chooses a historical analogy to strengthen his warning.

He references the lessons from the first phase of globalization: the entire industrial economy was hollowed out through outsourcing. "GDP figures look good on the surface, but the actual displacement is real, and the consequences are still unfolding."

His judgment is that AI is heading toward similar structural risks. If all value ultimately flows to a few foundational models, political and social tolerance will no longer be sustainable. "For an AI future that hollowed out entire industries, society will not grant permission."

This is a politically economic perspective warning: AI centralization is not just a business issue but a governance issue that could trigger social backlash. Nadella’s framework essentially defends a decentralized AI ecosystem—also a defense of Microsoft’s own cloud and enterprise AI strategies.

Microsoft’s business logic is clear: if companies build applications directly on OpenAI or Anthropic APIs, the intermediary value of Azure will be compressed. Nadella needs a narrative that convinces companies that "building their own AI capabilities" is more important than "renting models."

@E3#