An article written by Microsoft CEO Satya Nadella.

An article written by Microsoft CEO Satya Nadella:
This kind of article is truly worth reading—don’t look at most AI that’s written for the Chinese social media feed.

Satya:

I’ve been thinking about the future direction of companies in an AI-driven economic environment.

This transformation is fundamentally different from any platform change we’ve seen before. In the past, we used digital systems to enhance human capital. Now, for the first time, we’re able to build a real cognitive loop between humans and digital systems. This is refreshing, because it completely changes how we understand work inside enterprises.

The key isn’t about certain digital tools or systems and how we use them. Rather, in a world where AI models can keep absorbing professional knowledge from humans and organizations and turning it into something commodifiable, how do organizations continue to learn, build intellectual property, differentiate themselves, and thrive?

Every company has to build what I call human capital and token capital. Human capital includes employees’ knowledge, judgment, relationship networks, creativity, and pattern-recognition abilities, while token capital is the AI capabilities the company builds and owns.

What matters is that as token capital grows, the value of human capital doesn’t decrease—it only increases! I believe human initiative will be the driving force behind the growth of token capital. People will set ambitious goals, connect information across different domains, build relationship networks, and identify the most important patterns. Without people guiding, computers will just spin in place.
This means the real opportunity isn’t in choosing the best model—it’s in building a model-based learning loop that allows human capital and token capital to grow through compounding. You can outsource a task, or even a job, but you can never outsource learning. The future of enterprises depends on whether they can achieve compound growth of learning outcomes between people and AI.

This requires a completely new approach to architecture—one that enables every enterprise to build intelligent systems that improve over time while also maintaining control over its intellectual property. Enterprises should be able to replace existing “general” models without losing the “company veterans” expertise that’s built into their learning systems. This will be a crucial “test” of corporate control and autonomy in the future era.

Enterprises need to translate their own workflows, domain knowledge, and accumulated judgment into AI systems, and ensure those systems keep improving every time they’re used. Private evaluation should be able to capture whether the model has truly improved outcomes that are crucial to the business (not just external benchmarks!). A private reinforcement learning environment should allow the model to grow continuously based on real data inside the organization. Its knowledge base makes organizational memory queryable and improves token usage efficiency.

This loop will become the company’s new intellectual property. I compare it to a mountain-climbing machine. Unlike most assets, it has a compounding effect. Every improvement to the workflow produces better training signals, accelerating the accumulation of the company’s unique tacit knowledge. Companies that build this loop early will have advantages that are difficult to replicate, regardless of what new single-model capabilities they may possess.

The last thing we want to see is that all industries, all companies, hand their value over to a few models that grab everything. If all value is concentrated in the hands of a few models, the political and economic system absolutely won’t tolerate it. Society will never allow the future of AI to hollow out entire industries.

Think about what happened during the first phase of globalization: outsourcing drained entire industrial economies. On the surface, GDP data looked good, but industrial relocation was real, and its consequences are still showing up to this day. We must not let this pattern repeat in the AI era—where a small number of AI systems capture all the economic benefits while the entire industry watches its knowledge being commodified and ultimately destroyed.

I believe our top priority must be to build a cutting-edge ecosystem, not just a cutting-edge model—so that value can flow widely to every company, every industry, and every country. In this ecosystem, each organization can have a learning loop that encodes its institutional knowledge, continuously accumulating its human capital and token capital.

I’ve held this belief since I was young: platforms can create additional value that’s greater than what the platform itself provides, and every company can keep innovating to create its own value.

When this happens, companies won’t just create value for themselves—they’ll also create value for the surrounding economy. Employees’ expertise will be enhanced; their judgment will be integrated into replicable, scalable systems; and both enterprises and the surrounding communities will benefit.

This is how companies create value for themselves and for the wider economy. And this is the stable balance we should build together.