Your section on local governments reselling training data hints at something larger. In software AI, the intelligence layer commoditized fast: the most popular open-source agent hit https://thesynthesisai.substack.com/p/the-free-agent, with China already leading adoption. If embodied AI follows the same arc, value won't sit in the model or the robot. It sits in the physical loop: deployment data, factory access, supplier networks. Those Liuzhou data centers look less like training facilities and more like a wager that robotics intelligence commoditizes too, and whoever controls the proving ground keeps the margin.
Yes. This is what I describe as the I5 model, which provides a structured way of understanding how countries can go from idea to institutions and instruments and then infrastructure and impact with minimal loss of fidelity across stages.
China is the canonical case first with EV’s and now likely also with humanoid robots. The overlap between the EV supply chain and the robotic one adds another interesting dimension.
The model is fully elaborated in my book “The Innovation Machine: How China Creates and Adopts Technology Through Governance,” which was published by Springer Nature this year
Rui, Keep up the good work. Thanks for what you do to bridge these two countries and shed light on some important topics with on the ground perspectives. There’s way too much misinformation and misunderstanding of what’s really happening in the China tech scene today.
The future of work isn’t human.
And it’s not machine.
It’s both.
We’re entering the human–machine hybrid era:
AI coworkers Autonomous robots Smart factories AI-assisted healthcare & education
This isn’t automation 2.0.
It’s a redesign of how work actually happens.
Here’s the shift most people are missing:
You won’t compete against AI.
You’ll compete based on how well you work with it.
And the next big question?
Not “Will AI replace jobs?”
But:
Who becomes exponentially more powerful with it and who doesn’t?
Because by the 2030s, with AGI debates getting real
This won’t just reshape industries.
It will redefine what it means to be productive.
Your section on local governments reselling training data hints at something larger. In software AI, the intelligence layer commoditized fast: the most popular open-source agent hit https://thesynthesisai.substack.com/p/the-free-agent, with China already leading adoption. If embodied AI follows the same arc, value won't sit in the model or the robot. It sits in the physical loop: deployment data, factory access, supplier networks. Those Liuzhou data centers look less like training facilities and more like a wager that robotics intelligence commoditizes too, and whoever controls the proving ground keeps the margin.
Yes. This is what I describe as the I5 model, which provides a structured way of understanding how countries can go from idea to institutions and instruments and then infrastructure and impact with minimal loss of fidelity across stages.
China is the canonical case first with EV’s and now likely also with humanoid robots. The overlap between the EV supply chain and the robotic one adds another interesting dimension.
The model is fully elaborated in my book “The Innovation Machine: How China Creates and Adopts Technology Through Governance,” which was published by Springer Nature this year
Rui, Keep up the good work. Thanks for what you do to bridge these two countries and shed light on some important topics with on the ground perspectives. There’s way too much misinformation and misunderstanding of what’s really happening in the China tech scene today.
To be honest if I have the chance, I would like to explore China's tech scene for once and decide whether it's good to work on certain projects there.