Keywords: Capital warfare, technological sovereignty, “bathe-and-go” overseas listing, AI rivalry
I. Introduction: The “New Berlin Wall” Behind a Merger
In April 2026, what seemed like a routine cross‑border acquisition sent shockwaves through global tech circles.
China’s National Development and Reform Commission, citing national security, blocked U.S. tech giant Meta’s proposed acquisition of the Chinese AI agent start‑up Manus for over $2 billion. The ruling not only halted a major capital feast but also declared a new reality: in the age of artificial intelligence, capital flows must yield to strategic state security.
This is no isolated event. From Washington’s export controls to Beijing’s investment reviews, global capital is undergoing an unprecedented political stress test. Toward Chinese AI, global capital now holds a complex, ambivalent stance—viewing it as both an adversary to contain and an indispensable partner.
II. Defense and Counterattack: China Builds a “Capital Firewall”
The rejection of the Manus deal hinges on a regulatory principle: substance over form.
Although Manus had previously moved its headquarters to Singapore in a “bathe‑and‑go” overseas maneuver to bypass scrutiny, its core R&D team and data accumulation remained in China. For regulators, this was essentially the loss of a strategic Chinese AI asset. As experts have noted, Manus—as the world’s first general‑purpose AI agent with “operating system”‑level value—would have handed Meta a decisive edge in next‑generation human‑computer interaction.
China quickly followed up with more aggressive countermeasures. Beijing is now moving to restrict U.S. capital from entering its top tech companies, especially the so‑called “AI Four Little Dragons” and rising stars like DeepSeek.
This is a classic “pulling the firewood from under the pot” strategy. While the U.S. choked off chip supplies on the hardware side, China is now tightening the noose on the capital side—cutting off U.S. venture capital’s ability to harvest AI dividends or even siphon trade secrets from Chinese firms. What we are witnessing is an escalation from a “silicon war” to a capital war.
III. Inseparable Entanglement: The U.S. Market’s Dependence on China
But if one concludes that global capital will simply exit China, that misreads the depth of market integration.
On the very same day the Meta‑Manus deal was rejected, U.S.-listed LightSource Semiconductor—a Chinese company providing laser chips for AI data centers—reported earnings that shocked Wall Street: net profit soared 1,153%. In other words, no matter how loudly Washington proclaims “de‑risking,” the foundation of U.S. AI computing power still rests on critical hardware made in China.
An even deeper dependency lies in algorithms. Bloomberg data shows that Chinese AI models, led by DeepSeek and Tongyi Qianwen, are sweeping global markets with extreme cost‑effectiveness.
The cost of Chinese AI models is just 1/33 that of U.S. models; their global market share has surged from 1% to 15% in a single year. For profit‑driven global capital, unless they are prepared to forfeit the world’s largest application market and this uniquely efficient supply chain, “decoupling” will always be a money‑losing proposition.
IV. Water vs. Oil: The Clash of Two AI Philosophies
The root of this ambivalence lies in two fundamentally different conceptions of AI.
The U.S. treats AI as “oil” —a monopolizable, ownable commodity to be extracted for maximum rent. China treats AI as “water” —valuable precisely because it flows, functioning like ubiquitous public infrastructure.
While Silicon Valley debates the safety of large language models, China’s “city brains” are already optimizing traffic in Hangzhou, and Chinese industrial robots are reshaping global manufacturing. This pragmatic, infrastructure‑first approach leaves Western capital both fearful and perplexed.
They fear the speed of China’s AI ascent, yet feel helpless to stop it. As the Asia Times observed, the U.S. aims for a China‑free ecosystem, but capital, by its nature, chases the highest return on investment. And China remains the most compelling answer.
V. Conclusion: Toward a New Equilibrium
The collapse of the Meta‑Manus deal is not merely a business setback. It marks the birth pangs of a new global technology order.
For global capital, political correctness and commercial correctness are now tearing apart. Looking ahead, we will likely see more “two‑track” arrangements: U.S. capital nervously engaging with Chinese assets through indirect channels, even as it eagerly embraces China’s open‑source AI advances.
Chinese AI is moving from a defensive “stranglehold” posture to an active asset‑protection strategy. In this grand game, global capital will not abandon its reliance on China—but it must learn to abide by new rules. This is not just a contest of technologies; it is the emergence of a new sovereign logic in the age of AI.