In early 2025, Chinese regulators killed Meta’s attempt to acquire Manus, the AI startup behind an autonomous agent that had captivated the tech world. The National Development and Reform Commission did not negotiate or suggest alternative deal structures. It simply ordered all parties to walk away.
That decision now looks less like an isolated intervention and more like a preview. Beijing is preparing a formal approval system that would screen American capital flowing into China’s leading artificial intelligence and technology companies, according to a Bloomberg News report published in April 2026 that cited people familiar with internal government deliberations. If the rules take shape as described, U.S. investors would need government sign-off before putting money into Chinese AI firms, a requirement that does not currently exist.
Why the Manus block still matters
The clearest evidence of where Beijing is headed is not a leaked policy draft but a completed action. The NDRC’s Office of the Working Mechanism for Security Review of Foreign Investment, the body responsible for screening deals that touch national security, formally prohibited the Manus acquisition and required all parties to withdraw.
“This was Beijing drawing a red line in permanent marker,” said Paul Triolo, a senior vice president at the business advisory firm Albright Stonebridge Group, who tracks Chinese technology policy. “They picked a deal the whole world was watching and made sure the outcome was loud enough that no one could miss it.”
What made the decision striking was its precision. Regulators did not issue a sweeping directive covering an entire sector. They targeted one high-profile transaction between a U.S. tech giant and a Chinese AI company whose autonomous agent technology had drawn global attention. The message to the market was unambiguous: when Beijing decides a transaction threatens core technological interests, commercial logic on the buyer’s side counts for nothing.
The block also signaled something about timing. It came during a period when China was otherwise loosening conditions for foreign investors in its public markets, making the intervention all the more conspicuous.
What the reported new rules would change
China already operates a layered system for reviewing foreign investment. The State Council maintains an official foreign investment portal that aggregates regulations, sectoral negative lists, and departmental circulars from the Ministry of Commerce (MOFCOM) and the China Securities Regulatory Commission (CSRC). In late 2024, the State Council revised rules that lowered certain thresholds for overseas buyers taking strategic stakes in listed companies, a move widely interpreted as a welcome mat for foreign capital.
The reported new restrictions would carve out a pointed exception to that welcome mat. According to the Bloomberg report, American investors seeking to put money into Chinese AI and tech firms would need government approval before proceeding. The mechanism would run through MOFCOM and the CSRC, layered onto the existing review architecture rather than built from scratch.
Critical details remain undefined. As of May 2026, no draft regulation, public comment period, or State Council announcement has surfaced. Open questions include which sectors and subsectors would fall under the new screening, what ownership thresholds would trigger review, whether passive portfolio investments like index fund holdings would be captured, and whether the rules would apply retroactively to existing stakes.
Washington built the mirror image first
Beijing’s reported move did not emerge from nowhere. The United States has been constructing its own restrictions on the opposite side of the same corridor for years. In October 2023, President Biden signed an executive order directing the Treasury Department to regulate outbound U.S. investment in Chinese companies working on semiconductors, quantum computing, and artificial intelligence. The final rules took effect in January 2025, requiring American investors to notify the government of certain transactions and prohibiting others outright.
The symmetry is hard to miss. Washington screens American money going out. Beijing now moves to screen American money coming in. Each government frames its controls as defensive, grounded in national security rather than protectionism. But the practical effect is a narrowing channel for capital between the world’s two largest economies, with AI sitting at the most restricted point.
Neither government has publicly acknowledged the other’s moves as a direct trigger. The U.S. Treasury, Commerce Department, and White House have not commented on the reported Chinese plan. That silence makes it difficult to determine whether Washington views the prospective curbs as retaliatory, reciprocal, or simply the predictable result of a technology rivalry that has been escalating since at least 2018, when the first major semiconductor export controls landed.
The stakes for investors and startups
For American venture capital and private equity funds with active deal pipelines in Chinese AI, the most immediate concern is timing. Transactions currently in negotiation could face delays, revised terms, or outright rejection if new approval requirements take effect before closing. Even minority stakes may trigger review, particularly if the target company works on foundation models, large-scale compute infrastructure, or applications with potential military or surveillance uses.
Firms with existing holdings face a different set of risks. Many investment regimes distinguish between future and past transactions, but the scope of any grandfathering provisions is unknown. Investors holding positions in sensitive segments of China’s tech sector may need to stress-test scenarios in which exits become harder or governance rights are constrained by new regulatory conditions.
On the Chinese side, the effects are already visible. Startups that historically raised capital in U.S. dollars now face a more complicated fundraising environment. Even the perception of looming restrictions can freeze negotiations, as American funds hesitate to commit money that might land in regulatory limbo. That dynamic is pushing some Chinese AI companies toward renminbi-denominated funds, domestic stock listings, or partnerships with investors from countries that carry less political friction in Beijing’s eyes.
A secondary ripple could redirect capital flows entirely. If American investors encounter higher barriers in China, some portion of that money may shift toward AI ecosystems in South Korea, Japan, India, and the Gulf states, all of which are actively courting foreign investment in technology. But regulatory openness alone will not determine where the money goes. Access to engineering talent, computing resources, and local demand for AI products will weigh just as heavily.
Tracking the regulatory signals that will confirm or kill this plan
The Manus block is confirmed, public, and unambiguous. It remains the single strongest data point for anyone trying to understand how Beijing will treat American interest in Chinese AI assets. The broader plan for an approval-based screening system rests, for now, on a single Bloomberg News report citing unnamed sources. Bloomberg has a strong track record on early Chinese policy scoops, but the claim has not been independently corroborated by a second news organization or an official Chinese document. The Bloomberg report could not be independently verified because the outlet has not published a publicly accessible article URL for the story.
That gap matters. Until a draft rule surfaces on the State Council’s regulatory platform or MOFCOM issues new guidance, the reported restrictions should be treated as a credible signal of policy direction rather than a done deal. The practical takeaway for investors: prepare for tighter rules without assuming they are imminent or fully defined. Monitor official channels, track new circulars from MOFCOM and the CSRC, and treat the Manus decision as the clearest preview of what Beijing is willing to do when it decides a piece of its AI sector is too important to let go.
