Microsoft CEO: In the AI era, how do you define a company's moat?

Title: A frontier without an ecosystem is not stable
Author: Satya Nadella, Microsoft CEO
Translation: Peggy

Author: Rhythm BlockBeats

Source:

Reprint: Mars Finance

Editor's note: Microsoft CEO Satya Nadella believes that the true competitive advantage for enterprises in the AI era does not lie in betting on the strongest model, but in whether they can turn their workflows, domain knowledge, organizational judgment, and employee experience into a continuously evolving learning system. In other words, companies should not only purchase AI capabilities but also develop their own "learning closed loop" (a system where human experience, business processes, and model capabilities continually reinforce each other).

Under this framework, future companies will accumulate two types of capital: human capital, which includes employees' knowledge, judgment, network relationships, creativity, and pattern recognition abilities; and Token Capital (AI capabilities built and owned by the enterprise itself). Nadella emphasizes that AI will not devalue human capital; instead, it will make goal setting, cross-domain connections, and key pattern recognition more important. Without human direction, computing power just spins in place; without organizational knowledge accumulation, even the strongest models are just external tools.

The core judgment of this article is: a frontier without ecosystem support will not be a stable future. The value of AI should not be swallowed by a few general models but should form a frontier ecosystem, enabling every company, industry, and country to have its own learning closed loop. Enterprises need to establish private evaluation, private reinforcement learning environments, and queryable knowledge bases to turn tacit experience into reusable, scalable, and iterative system capabilities. The true moat may not be a specific model itself but the ability to retain "company veteran" experience even after replacing general models.

This is also the key to enterprise sovereignty in the AI era: whoever can turn organizational knowledge into a continuously compounding system will be able to retain IP, amplify employee capabilities, and keep the economic value brought by AI within their own business, industry, and community.

Below is the original text:

I have been thinking recently about what the future of enterprises will look like in an AI-driven economy.

This transformation is unlike any previous platform migration. In the past, we used digital systems to enhance human capital; this time, it’s the first opportunity for us to establish a true cognitive closed loop between humans and digital systems. This is profoundly disruptive because it will change how we understand "work" within enterprises.

The real key issue is not how a digital tool or system is used, but how, in a world where AI models can continuously absorb human and organizational expertise and commodify it, organizations can continue to learn, accumulate intellectual property, differentiate themselves, and thrive.

Every company must build what I call human capital and Token Capital. Human capital includes employees’ knowledge, judgment, network relationships, creativity, and pattern recognition; Token Capital, on the other hand, is AI capability that the enterprise constructs and owns itself.

Importantly, as Token Capital grows, human capital does not become less important. Quite the opposite, it becomes even more critical. I believe human agency will be the core driver of Token Capital growth. Humans will set ambitious goals, connect clues across domains, build relationships, and identify truly important patterns. Without human direction, computing power will just spin in place.

This means that real opportunity does not lie in choosing the best model but in building a learning closed loop on top of models, allowing human and Token Capital to grow mutually and exponentially. You can outsource a task, even outsource a job, but you can never outsource your learning. The future of enterprises depends on whether this learning can continue to compound between humans and AI.

This requires a new architectural approach: every enterprise should be able to build intelligent agent systems that improve over time while maintaining control over their own intellectual property. A company should be able to replace a "generalist" model without losing the "veteran" expertise embedded in its learning system. This will be a key test of future enterprise control and sovereignty.

Enterprises need to transform their workflows, domain knowledge, and long-term judgment into AI systems that continuously improve with each use. Private evaluation should measure whether models truly improve on business outcomes that matter to the enterprise, not just external benchmarks. Private reinforcement learning environments should enable models to get stronger based on real organizational trajectories. Knowledge bases will turn institutional memory into queryable assets and improve Token efficiency.

This closed loop will become a new form of intellectual property for enterprises. I see it as a "climbing machine." Unlike most assets, it will grow through compound interest. Every workflow improvement generates better training signals, accelerating the accumulation of tacit knowledge unique to the enterprise. Companies that establish this system early will gain an irreplicable advantage, regardless of breakthroughs in individual model capabilities in the future.

What we least want to see is a world where every industry’s value is handed over to a few models that devour all content. If all value is ultimately captured by a handful of models, political and economic structures will not tolerate such an outcome. A future where AI hollowed out entire industries cannot gain societal approval.

Think back to the first phase of globalization: entire industrial economies were outsourced and hollowed out. On the surface, GDP figures looked decent, but real industry shifts and employment impacts were significant, and their consequences are still felt today. We cannot bring this dynamic into the AI era—allowing a few AI systems to capture all economic returns while industry knowledge is commodified and drained beneath them.

In my view, our priority must be to build a frontier ecosystem, not just a frontier model. Only then can value flow broadly to every company, industry, and country. In such an ecosystem, each organization can develop its own learning closed loop, encode its institutional knowledge, and allow human and Token Capital to grow together exponentially.

This aligns with my long-standing belief in platform principles: the value created on a platform should exceed the value captured by the platform itself; every company should be able to innovate continuously and create its own value.

When this is achieved, enterprises will generate value for themselves and for the broader economy. Employees’ expertise will be amplified, their judgment integrated into systems, and these benefits will be scalable and replicable, flowing back into the company and its surrounding community.

This is the way enterprises create value for themselves and the wider economy. It is also the stable, balanced system we should jointly build.