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 single strongest model, but in whether they can distill their workflows, domain knowledge, organizational judgment, and employee experience into a continuously evolving learning system. In other words, companies cannot just purchase AI capabilities; they must own their own "learning loop" (a system where human experience, business processes, and model capabilities constantly reinforce each other).
Within this framework, future companies will simultaneously accumulate two types of capital: Human Capital, which includes employees' knowledge, judgment, relationship networks, creativity, and pattern recognition abilities; and Token Capital (the AI capabilities that a company builds and owns itself). Nadella emphasizes that AI will not devalue human capital; on the contrary, it will make human capabilities in goal-setting, cross-domain connection, and key pattern recognition more important. Without human direction, computing power just spins its wheels; without an organization's own knowledge base, even the strongest model is just an external tool.
The core judgment of this article is: A frontier without an ecosystem will not be a stable future. The value of AI should not be swallowed by a few general-purpose models but should form a frontier ecosystem, allowing every company, every industry, and every country to own its learning loop. Enterprises need to establish private evaluations, private reinforcement learning environments, and queryable knowledge bases, turning tacit experience into reusable, scalable, and iterable system capabilities. The true moat may not be a specific model itself, but rather the "company veteran"-style experience an enterprise would not lose even after switching out the underlying general model.
This is also key to corporate sovereignty in the AI era: Those who can turn organizational knowledge into a system that generates continuous compound interest will retain intellectual property, amplify employee capabilities amidst rapid model iteration, and keep the economic value brought by AI within their own business, industry, and community.
Here is the original text:
I've been thinking a lot lately about what the future of the enterprise will look like in an AI-driven economy.
This transition is different from any previous platform shift. In the past, we used digital systems to augment human capital. This time, it's the first time we can establish a true cognitive loop between people and digital systems. This is very disruptive because it changes how we understand the very nature of "work" inside an enterprise.
The truly critical question is not how a particular digital tool or system is used, but rather how an organization continues to learn, accumulate intellectual property, differentiate itself, and thrive in a world where AI models can continuously absorb human and organizational expertise and productize it.
Every company must build what I call Human Capital and Token Capital. Human Capital includes employees' knowledge, judgment, relationship networks, creativity, and pattern recognition. Token Capital is the AI capability that the enterprise itself builds and owns.
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 set ambitious goals, connect dots across domains, build relationships, and identify the patterns that truly matter. Without the direction of human purpose, compute just churns.
This means the real opportunity is not in choosing the best model, but in building a learning loop on top of the model that allows Human Capital and Token Capital to compound each other. You can outsource a task, you can even outsource a job, but you can never outsource your learning. The future of the enterprise lies in whether it can make this learning compound continuously between people and AI.
This requires a new architectural mindset: Every enterprise should be able to build agentic systems that improve over time, while still retaining control over their intellectual property. A company should be able to swap out a "generalist" model without losing the "company veteran" expertise ingrained in its learning system. This will be a key test of control and sovereignty for the future enterprise.
Enterprises need to transform their workflows, domain knowledge, and accumulated long-term judgment into AI systems that get better with every use. Private evaluations should measure whether a model is actually improving on business outcomes the company cares about, not just external benchmarks. Private reinforcement learning environments should make models stronger based on real internal trajectories. Corporate knowledge bases should make institutional memory queryable and improve token efficiency.
This closed loop becomes the new intellectual property of the enterprise. I think of it as a "hill-climbing machine." And unlike most assets, it compounds. Every workflow improvement yields a better training signal, accelerating the accumulation of a company's unique tacit knowledge. Companies that build this system earlier will gain a hard-to-replicate advantage, regardless of how individual model capabilities break through in the future.
What we least want to see is a world where companies across every industry cede their value to a handful of models that ingest everything they see. If all value ultimately accrues to a few models, the political economy simply will not tolerate it. An AI future that hollows out entire industries will not gain a social license.
Think about what happened in the first phase of globalization: entire industrial economies were hollowed out by outsourcing. On the surface, GDP numbers looked okay, but the real shifts in industry and employment shocks occurred, and their consequences are still felt today. We cannot let this dynamic play out in the AI era—where a handful of AI systems capture all the economic returns while the knowledge of entire industries is commodified and hollowed out beneath them.
In my view, our priority must be to build a frontier *ecosystem*, not just a frontier *model*. Only then can value flow widely to every company, every industry, every country. In such an ecosystem, every organization can own its learning loop, encode its institutional knowledge into it, and compound its Human Capital alongside its Token Capital.
This is also the platform ethos I have always believed in: The value created on the platform should be greater than the value captured by the platform; every company should be able to innovate continuously and create its own value.
When that happens, enterprises will create value for themselves and for the economy they operate in. Employees' professional capabilities will be amplified; their judgment will become part of the system, replicable and scalable, and those returns will flow back to the company and its surrounding community.
That is how enterprises create value for themselves and the broader economy. And it is the stable equilibrium we should all build toward.
