And the American dollar fund buddies talked again about the Manus acquisition case.

This is a very worthwhile case to analyze in detail; it shouldn't be viewed only from the technical route and product capabilities.
The real focus is that it collided with the reverse version of CFIUS logic, Reverse CFIUS.
In the past, everyone understood CFIUS as a tool used by the U.S. to review foreign investment entering the U.S.
Who wants to buy American companies, American chip assets, American data companies, after you come in, will it threaten national security?
CFIUS is about who can enter the U.S., and the Manus incident exposed what cannot leave China this time.
Manus originally followed a familiar path: Chinese teams develop the product, Singapore hosts the main entity, U.S. capital flows in, and finally sells to major American firms.
This approach was established in the SaaS era, but in the AI Agent era, the logic has changed.
Because Manus is not an ordinary software; it connects to user workflows, behavioral data, task chains, and decision entry points.
So regulators are not looking at where it is registered, but where the technology comes from?
Where does the team originate?
Where do the data and model capabilities come from?
After being acquired by Meta, where will these capabilities flow?
In other words, the registration location cannot erase the source of technology, the ownership structure cannot erase the team origin, and capital pathways cannot erase sovereignty—these are all impossible to explain away.
The registration location arbitrage may no longer work today; AI companies face "technological sovereignty review."
The U.S. uses CFIUS to regulate foreign investment, and China has started using Reverse CFIUS to control the outflow of technology.
In the future, when AI companies go overseas, the most important question will no longer be where you are registered, but who truly owns your technology.