China to curb U.S. funding for local tech firms after Meta-Manus deal
For years, Chinese AI startup founders chased Silicon Valley money as a badge of credibility and a fast track to global scale. That era may be over. Beijing has ordered Chinese AI startups and private tech companies to reject American investment without prior government approval, a sweeping directive that represents the sharpest escalation yet in the regulatory fallout from Meta’s acquisition of the artificial intelligence firm Manus.
The instruction came from officials at the National Development and Reform Commission, China’s top economic planning body, in a directive delivered in April 2026, according to Bloomberg News, which cited people familiar with the matter. The NDRC has not published the directive as a formal regulation, but its involvement signals the restriction carries authority across the Chinese government. For the thousands of startups that have historically courted venture capital from firms like Sequoia, Andreessen Horowitz, and other U.S. investors, the message is blunt: no American dollars without Beijing’s sign-off.
The Meta-Manus deal lit the fuse
Manus, a Chinese-founded AI company that built tools for autonomous digital agents capable of executing complex tasks across software platforms, caught Meta’s attention as the social media giant raced to embed AI across its products. The acquisition’s exact terms have not been disclosed, but the deal quickly drew scrutiny in Beijing, where officials viewed it as a case study in how strategically valuable Chinese-origin AI technology could be absorbed by a foreign buyer.
China’s Ministry of Commerce has opened a formal investigation into the purchase, probing the transaction across four fronts: export controls, technology transfer, outbound investment, and cross-border mergers and acquisitions. MOFCOM spokesperson He Yadong outlined those grounds to reporters, as the Associated Press reported. The breadth of the review signals that Beijing treats the deal not as a one-off transaction but as a precedent that could shape how all future cross-border AI acquisitions are handled.
Enforcement has already moved beyond paperwork. Chinese authorities imposed travel and exit restrictions on Manus executives, The Washington Post reported (note: the specific Washington Post article URL could not be confirmed beyond the outlet’s homepage), even though the company had previously relocated its corporate headquarters outside China. That detail carries outsized implications: Beijing is asserting jurisdiction over personnel tied to a Chinese-founded firm regardless of where it is now incorporated. For any startup founder who restructured abroad to attract foreign capital, the precedent is chilling.
The legal foundation for these moves predates the current dispute. In August 2020, MOFCOM and the Ministry of Science and Technology jointly updated the Catalogue of Technologies Prohibited or Restricted from Export through Announcement No. 38, adding AI-related provisions that give regulators a ready-made framework for blocking or conditioning deals involving sensitive technology. The current probe appears to draw on those existing rules rather than on new legislation drafted for the occasion.
Critical details remain unresolved
The NDRC’s directive has not been published in any official gazette, leaving the precise scope, duration, and enforcement mechanism unavailable for startups or their lawyers to read. How many companies have received the instruction, and how many active funding rounds have stalled as a result, is not publicly known. Meta, U.S. venture firms with exposure to Chinese AI, and affected startups have not publicly commented.
The specific technologies at issue in the Meta-Manus probe are also opaque. Neither MOFCOM nor He Yadong publicly identified which AI capabilities or datasets triggered the investigation, making it difficult to judge whether regulators are targeting a narrow set of proprietary algorithms, training data sourced in China, or a broader category of AI research that could sweep in other cross-border deals.
The travel restrictions raise their own questions. Reporting confirms the bans exist and are tied to the probe, but the legal basis for applying them after a corporate relocation has not been publicly detailed. Whether the restrictions extend beyond the executive level, or to former employees who left before the acquisition, is unclear.
In China’s fragmented regulatory environment, national directives can land unevenly. Some provincial authorities may apply the NDRC’s guidance narrowly, focusing on firms with direct ties to sensitive military or intelligence applications. Others could read it expansively enough to freeze a much wider range of cross-border deals. Without a published rulebook, companies are left navigating the ambiguity through private consultations and educated guesses.
A tightening vise from both sides
China’s new restrictions do not exist in isolation. The United States has been building its own barriers to AI-related capital flows in the opposite direction. Executive Order 14105, signed by President Biden in August 2023 and implemented through Treasury Department rules that took effect in early 2025, requires U.S. persons to notify the government of, or in some cases refrain from, investments in Chinese companies working on advanced AI, semiconductors, and quantum computing. The result is a closing loop: Washington screens money going out, and Beijing now screens money coming in.
For American venture capital and growth equity firms, the practical effect is a shrinking map. U.S. investors in Chinese AI already faced compliance obligations under the outbound investment rules. Now they face the additional hurdle of needing approval from Chinese regulators who have every incentive to use the review process as leverage. Deals that once closed on the strength of a term sheet and a wire transfer may now require months of parallel regulatory review in two capitals with competing strategic interests.
The financial stakes are substantial. U.S. venture firms have poured billions of dollars into Chinese AI companies over the past decade, funding everything from large language model developers to autonomous driving startups. While that flow had already slowed under pressure from both governments, the NDRC directive threatens to formalize a barrier that was previously a matter of political risk rather than regulatory prohibition.
One open question is who fills the gap. Chinese state-backed funds, sovereign wealth vehicles from the Gulf states, and domestic private equity firms could absorb some of the demand for capital. But U.S. venture money has historically come bundled with technical networks, talent pipelines, and global distribution channels that are harder to replicate. Losing access to that ecosystem, not just the dollars, could slow the international ambitions of Chinese AI firms even as their domestic capabilities continue to advance.
Beijing holds the key to cross-border AI capital
The safest reading of this moment is that China is stress-testing a more interventionist posture on who can fund and control the companies building its most strategically important technology. The Meta-Manus case serves as both the catalyst and the warning shot. The combination of an authoritative but unpublished NDRC directive, a formal MOFCOM probe grounded in existing export-control law, and concrete restrictions on individual executives suggests Beijing is prepared to police not just where sensitive AI ends up, but who writes the checks that keep it moving.
For founders weighing a U.S. term sheet against a domestic or non-American offer, the calculus has shifted overnight. For American investors eyeing Chinese AI, the door has not closed, but a new lock has been installed and Beijing holds the key. As of May 2026, with the underlying rules still unpublished and untested, everyone in this market is operating with partial information and rising stakes.
