I found Satya Nadella’s speech quite interesting.
He said this: “What if we could make use of the CPUs, GPUs, and NPUs inside countless PCs around the world?” And he also gave the reason: “To solve the bottlenecks in AI data centers.”
Many people will hear this and think of technological innovation. Connecting computing resources from around the world to build more powerful AI. Perhaps it could even help solve some of humanity’s biggest problems. It’s certainly an appealing idea.
But I had a somewhat different thought.
“So we’re being asked to become voluntary node-slaves for a centralized AI server?”
Of course, that expression may sound extreme. But the question I want to ask is a different one.
Computation is being decentralized, but is power being decentralized as well?
Costs are being distributed, but is control being distributed too?
Participation is being decentralized, but is decision-making authority being decentralized?
I already operate a Zebra node on a blockchain network. I know what it means to participate in a network. I’m not just standing on the sidelines. I actively contribute to maintaining the network. Because of that, I do not automatically reject distributed systems. If anything, I understand their value better than most people.
But at the same time, I also know this:
Networks can have a powerful impact on human society, both for good and for bad.
Human beings are social creatures. When someone says they have discovered something remarkable, people naturally want to be included in that network.
“Wow, that’s incredible.”
“I want to be part of that group.”
“I want to be part of that project.”
These feelings are natural. They may even be one of the driving forces behind human progress.
The problem comes after that.
We often mistake participation and belonging for a sharing of power.
But participating in a network and controlling a network are completely different things.
Whether we’re talking about blockchain or AI, the important questions are ultimately the same.
Who makes the rules?
Who decides the updates?
Who can see the data?
Who can leave the system?
And who has veto power?
The future Satya Nadella describes is clearly a transition from an era where everything was processed centrally to an era of distributed computing. It is difficult to deny that direction itself. In fact, AI and cloud computing are already moving in that direction.
But distributed computation does not necessarily mean distributed power.
Billions of PCs may participate in the network. Billions of people may contribute to maintaining the system. But if policy decisions, model updates, identity systems, and service access rights remain centralized, then what exactly have we decentralized?
The computation?
The costs?
Or the responsibility?
I do not look at corporations through the lens of good and evil.
Corporations are neither angels nor demons.
Corporations follow incentives.
What generates profit?
What reduces costs?
What creates a competitive advantage?
I believe that is the most realistic way to understand corporate behavior.
That is why I do not interpret the future based on assumptions of corporate goodwill. Not because I refuse to believe in goodwill, but because goodwill itself cannot be analyzed.
Instead, I look at structure.
I look at the structure of technology, the flow of data, and where power is concentrated.
Because the structure of technology and the structure of data eventually shape the structure of society.
As people become more familiar with blockchain, AI agents, and distributed computing, they will gradually become more accustomed to living as part of a network.
And that may not simply be a technological change. It may be the beginning of a social change.
Perhaps future human beings will no longer stand at the center of production in the way they once did.
Instead, they may become participants, contributors, validators, and connectors across countless networks.
I am not saying that this is necessarily a good future.
Nor am I saying that it is necessarily a bad future.
But one thing is clear.
It is dangerous to assume that distributed computation automatically means distributed power.
And whenever a new technology appears, the first question we should ask is not how impressive the technology is.
The question should be:
“When this system is fully built, where does the power reside?”