China blocks Meta's $2 billion purchase of Manus, chilling the AI market

China's decision to block the sale of Manus to Meta for $2 billion is more than the cancellation of one big deal. It is a direct signal to the entire AI market: in 2026, the fate of strategic technologies is determined not only by money, product and growth rates, but also by state control over who gets key artificial intelligence developers.

The deal looked emblematic. Meta, which has long been ramping up its AI investment, wanted to buy Manus, a startup working in agentic AI. These are systems that do not stop at answering a query: they can plan steps, carry out chains of actions and work as digital operators. For Meta, the purchase could have been a way to speed up its own AI products and strengthen its position in one of the most talked-about areas of the market. For Manus, it would have been a fast and lucrative exit that confirmed the high valuation of Chinese teams in the global race.

But the Chinese authorities decided otherwise. Blocking the deal means Beijing is not ready to freely let notable AI assets pass under the control of American corporations, especially when the technologies in question may be considered strategic.

This is an important signal for startups, investors and buyers. Until now, many young companies built their plans on the standard logic of the tech market: grow, prove the product's value, then either raise another round or sell to a big player. In Chinese AI, that path is now far less straightforward.

The timing is especially sensitive. The deal was cancelled a few weeks before a high-profile meeting between US President Donald Trump and Chinese President Xi Jinping. Against that backdrop, any cross-border AI deal automatically moves beyond corporate negotiations. It comes to be seen as a question of technological sovereignty, control over talent and the long-term balance of power between the world's two largest economies. Even if the decision is formally framed as regulatory, the market will almost inevitably read it as a political gesture.

For China's AI ecosystem, this has a chilling effect for several reasons. First, exit predictability falls: selling to an American tech giant no longer looks like a reliable end game, even at a high valuation. Second, investors will price companies more cautiously if they understand that the pool of potential buyers is shrinking because of political restrictions. Third, startups themselves will have to plan ahead for growing in a more closed environment: finding local partners, relying on domestic capital or adjusting their business structure to regulators' requirements. For an industry where speed often decides everything, this is a serious complication.

Meta also gets an unpleasant lesson. Money and strategic interest are no longer enough for big platforms to buy a promising AI asset. If development takes place in a country where artificial intelligence has become part of national technology policy, any takeover attempt can turn into a diplomatically sensitive case. That means large players will have to either bet on in-house research or look for more complex forms of cooperation, from partnerships to licensing, instead of outright purchases.

The main takeaway is that the global AI market is splitting along political lines ever faster. Capital, talent and technology are still global, but the ability to move them freely across borders is becoming the exception rather than the rule. The Manus story shows that for AI companies, it now matters not only how strong their product is, but also which jurisdiction they operate in, who may own them and how their growth fits the strategic interests of the state. For the market, this is bad news for freedom of dealmaking, but a very precise reminder that AI has finally become a matter of high politics.

*Meta has been designated an extremist organization and is banned in Russia.