According to information reported by Reuters, Washington is reportedly considering asking dozens of countries to choose between the American tech ecosystem and the Chinese one, with some partners potentially being warned that cooperation with Chinese initiatives would jeopardise their participation in a US-backed technology coalition. On 19 August, Beijing responded by calling for respect for each country's digital sovereignty.
We have covered the industrial dimension of this rivalry, from export controls to compute sovereignty. Here is its shift into the diplomatic dimension, and the change in nature is worth understanding.
What would change
Until now, restrictions have targeted flows: which chips can go where, which technologies can be exported. A country could buy American on some matters and Chinese on others.
A demand to pick a side targets alignments. It turns a technical and economic trade-off, made case by case, into an overall diplomatic position.
For many countries, this is a real constraint. A state may want American chips for its data centres, Chinese equipment for its networks, and open models from both. These choices today come down to performance and price. Tomorrow they would come down to positioning.
The argument on each side
On the American side, the reasoning is about security. Critical infrastructure relying on components and models controlled by a strategic rival creates dependency. This is the very argument we have made for Europe regarding open models: relying entirely on others for an essential technology is a risk.
On the Chinese side, the appeal to digital sovereignty consists of turning that argument around: each country should be free to choose its partners according to its own needs. This is an effective argument with states that do not want to depend on Washington either.
Both positions invoke the same value, independence, and draw opposite conclusions from it. That is what makes this debate hard to settle from the outside.
Europe finds itself in the most delicate situation. It has a solid regulatory framework, credible players in open models, and critical links in the semiconductor chain, particularly in manufacturing equipment without which no advanced chip is produced anywhere. This last point gives it real leverage. What it lacks is the capacity to produce AI chips at scale on its own soil. Leverage is not autonomy, and a demand to pick a side would make that difference very concrete.
What it reveals about the moment
This development confirms what we have been observing for weeks: the AI competition has left the realm of capabilities for that of public power.
We saw it with a state publishing its own open model, with labs hiring senior jurists, with decrees regulating data centres. What is decided in ministries now weighs as much as what is decided in labs.
What to take away
Nothing is set in stone: this is a reported direction, and demands of this kind are negotiated at length before producing any effect.
But the direction is clear, and it has a practical consequence for everyone. If the ecosystems split apart, portability becomes a key issue: being able to switch suppliers, self-host, not depend on a single infrastructure. These are exactly the recommendations we have been making for months on commercial grounds. They could become geopolitical recommendations.
There is an irony worth noting to finish. A world where every country must choose its technological camp is also a world where open models take on new value: they are the only objects that cross borders without authorisation, and that no one can take back once downloaded.