The story of China’s internet is not just about censorship or control. It is about tension. A constant back and forth between innovation and authority. That same tension is now shaping the future of artificial intelligence.
To understand where AI is headed, it helps to go back to where China’s digital journey began.
From 600,000 Users to a Digital Superpower
In 1997, when the term “Great Firewall” first appeared, China’s internet was tiny. Just a few hundred thousand users. The country was opening up economically, and the internet felt like a gateway to the world.
But even in those early days, there were signals that this openness would come with limits.
- The state never fully stepped away
- Red lines were never clearly defined
- Users learned boundaries through trial and error
This created a unique system. Not fully free. Not fully controlled. But constantly evolving.
A Digital Culture Built on Push and Pull
Journalist Yi-Ling Liu describes China’s internet as a dance between censor and user. That description captures something important.
People found ways to express, connect and build. Even within constraints.
Take the example of Ma Baoli.
- A police officer who discovered his identity online
- Built Blued, one of the world’s largest gay dating platforms
- Grew it into a massive business with millions of users
This was the promise of the internet. Discovery. Freedom. Opportunity.
But that promise had limits.
- Rising regulation
- Cultural tightening
- Political oversight increasing over time
Eventually, even successful entrepreneurs like Ma faced pressure. Some stepped down. Others disappeared from public view.
The system allowed growth. But only up to a point.
That Same Pattern Is Now Playing Out in AI
Fast forward to today, and AI is the new frontier.
The dynamics look familiar.
When DeepSeek launched its R1 model in 2025, it shocked the global tech ecosystem.
- Comparable performance to top US models
- Built at a fraction of the cost
- Less reliance on cutting edge chips
The reaction was immediate.
- Investors reassessed valuations
- Policymakers saw strategic risk
- The AI race became more geopolitical
But the bigger story is not competition. It is structure.
China’s Open Source Strategy Is Not What It Seems
At first glance, China’s approach to AI looks surprisingly open.
- Open source models
- Shared datasets and tools
- Encouraging developer ecosystems
Even robotics companies are opening up parts of their stack.
This seems counterintuitive for an authoritarian system.
But it follows a familiar logic.
- Openness is encouraged when it accelerates growth
- It is tolerated when it serves national goals
- It is restricted when it threatens control
There are already clear signs of this balance.
- Platforms like Hugging Face reportedly blocked
- Research pointing to political bias in Chinese AI models
- Restrictions on tools like autonomous AI agents in sensitive sectors
So while the surface looks open, the boundaries are still very real.
The Rise of “Walled Garden AI”
What is emerging is not fully open AI. And not fully closed either.
It is something in between.
- Open development within controlled ecosystems
- Innovation guided by state priorities
- Access shaped by ideology and policy
This is very similar to how China’s internet evolved.
What started as a global, open system gradually became:
- Closed
- Siloed
- Highly optimized for domestic control
AI could follow the same path.
Why This Matters Beyond China
This is not just China’s story. It is a preview.
As AI becomes more powerful, governments everywhere are asking similar questions.
- How much openness is too much
- Where should control begin
- Who decides what AI can say or do
Even in more open systems, we are already seeing:
- Debates around regulation
- Concerns about misinformation
- Questions about data sovereignty
The balance between innovation and control is becoming a global issue.
The Core Tension Will Define the Future
The biggest takeaway is simple.
Technology does not evolve in isolation. It evolves within power structures.
China’s experience shows:
- Innovation can thrive under constraints
- But those constraints eventually shape outcomes
- And sometimes limit the very systems they helped build
AI today is at that early internet stage.
Full of possibility. Rapidly evolving. Still undefined.
But the same question remains.
How far will openness be allowed to go before control pulls it back?