The Manus Acquisition Blocked: Where Are the Legal Red Lines for AI Technology Flight?

On April 27, 2026, China's National Development and Reform Commission (NDRC) officially blocked Meta's $2 billion acquisition of Manus — the first publicly halted AI-sector foreign acquisition since the implementation of the Measures for Security Review of Foreign Investment in 2020. From viral fame to offshore relocation, from a blockbuster acquisition to regulatory intervention, Manus's story reflects the deep contradictions facing China's AI industry amid globalization.

Manus's Rise: From Viral Sensation to Controversy

On March 6, 2025, Manus burst onto the scene as the "world's first general-purpose AI Agent." Unlike traditional AI assistants, it could autonomously complete complex tasks within an independent virtual machine — screening resumes, analyzing real estate, researching stock correlations, writing Python code — delivering final results for work that previously took humans hours.

The product launch instantly went viral on the Chinese internet. Invitation codes were resold for 100,000 RMB on secondary markets, and within two days its WeChat Index surpassed DeepSeek, making it the hottest AI product at the time. However, controversy followed the hype. Some users reported unstable performance and unclear capability boundaries. More critically, Manus had no proprietary foundation model — it was essentially a "wrapper" around others' models. Founder Xiao Hong compared this to "a consumer electronics company using someone else's chips," but the explanation failed to quell skepticism, and a significant gap emerged between product reputation and market expectations.

Capital-Driven De-Sinicization

Just three months after launch, Manus made a decision that shook the industry: relocating its headquarters to Singapore and undertaking a thorough de-Sinicization.

In June 2025, Manus's operating entity was changed to a Singapore company, Butterfly Effect PTE. Of the 120-person domestic team, only 40 relocated to Singapore; the remaining 80 were laid off (with compensation packages of N+3 or even 2N). All domestic social media accounts were wiped clean, and the official website blocked Chinese IP addresses.

On the surface, this appeared to be a "globalization strategy," but the real driver was capital pressure. In April 2025, Manus had just completed a $75 million Series B round led by Silicon Valley's Benchmark, with its valuation skyrocketing from $85 million to $500 million. However, the funding immediately attracted scrutiny from the U.S. Treasury Department's Outbound Investment Security Program. Benchmark and other investors directly demanded that Manus relocate its headquarters out of China. The so-called "going global" was, in reality, a forced move under pressure from American capital.

Meta's Blockbuster Acquisition and Regulatory Intervention

On December 30, 2025, Meta announced its $2 billion acquisition of Manus. The founder would serve as Meta Vice President reporting directly to the COO — Meta's third-largest acquisition in history. The news triggered a polarized domestic reaction: some saw it as a shining moment for Chinese AI entrepreneurs, while many more felt uneasy — a company whose core technology was developed in China had become an American corporate asset simply by routing through Singapore.

The Chinese government responded swiftly. On January 8, 2026, a Ministry of Commerce spokesperson publicly stated that the transaction would be investigated in accordance with the law. By March, media reports revealed that co-founder Xiao Hong and Chief Scientist Ji Yichao had been placed under exit bans, escalating the review from an economic matter to a national security concern. On April 27, the NDRC formally issued its prohibition order, requiring all parties to unwind the acquisition.

Registration Jurisdiction Is Not a Shield: Legal Logic Explained

The most critical legal question in this case is: On what basis can the Chinese government regulate a company registered in Singapore?

The answer is clear: regulators don't look at where a company is ultimately registered. They look at how the core team, R&D capabilities, training data, and intellectual property were originally transferred out of China. As long as the transfer chain passes through China, Chinese law has jurisdiction.

Manus crossed at least two red lines:

Technology Export Control Red Line

Its core AI technology likely falls under the information processing technology category in the Catalog of Technologies Prohibited or Restricted from Export. If the technology was transferred to an overseas company through personnel relocation, code sharing, or business migration without obtaining an export license, this directly violates technology export regulations.

Data Cross-Border Compliance Red Line

The product's training process used large volumes of Chinese user data. If this included personal information that was transmitted to an overseas company without going through proper compliance procedures, it violates data cross-border transfer laws.

In other words, re-registering in Singapore merely changed the "shell" — the core assets were never properly cleared for export.

Manus vs. DeepSeek: Two Starkly Different Paths

Intriguingly, on the same day the NDRC blocked Manus, DeepSeek released a new-generation fully open-source large model, following a path of domestically-produced computing power, ultra-low-cost inference, and full-stack autonomous control — every keyword precisely aligned with national policy priorities. The result: DeepSeek not only received policy support but was also embraced by capital markets, with its valuation continuing to climb.

The outcomes of these two paths form a stark contrast: autonomous control and compliant operations represent the main track, while "laundering-style going global" is a dead end.

A Casualty of Geopolitical Competition

The deeper context of the Manus case is the ongoing escalation of US-China tech competition. The U.S. has activated a "reverse CFIUS" mechanism, strictly limiting American capital flows into China's AI, semiconductor, quantum, and other frontier sectors. China, meanwhile, has continuously refined its foreign investment security review and technology export management systems. Manus happened to collide with this inflection point — American capital forced its de-Sinicization, while Chinese regulators decisively blocked the technology transfer. This was not a normal case of a company going global, but rather a high-risk probe in a complex geopolitical environment — its shutdown was virtually inevitable.

Takeaway: Compliance Is the Only Passport for AI Going Global

This case sounds an alarm for Chinese tech entrepreneurs. The NDRC's prohibition order marks a watershed moment, signaling that China's regulatory approach to critical technology sectors has shifted from reactive response to proactive rule-setting. It draws a clear legal bottom line: going global is fine, but packaging core technology nurtured in China and selling it to foreign companies through "laundering-style globalization" is absolutely unacceptable.

Regardless of where a company is registered or how its capital structure changes, if the technology is fundamentally Chinese, it must comply with Chinese law. For all AI startups planning capital operations, compliance costs have become an inescapable hard constraint. Only by promoting the orderly flow of technology and capital within a compliance framework can companies move more steadily and further in global markets.