Late last year, Meta Platforms announced what its executives framed as a decisive move in the global race for artificial intelligence. The company had acquired Manus, a Singapore-based startup that had built a general-purpose AI agent capable of writing code, conducting market research, managing files, and completing dozens of other complex tasks with minimal human direction. The price was over $2 billion. The deal, Meta said, would anchor a new division it called Superintelligence Labs.
Three months later, the two men who built Manus cannot leave China.
On March 25, the Financial Times and Reuters reported that Xiao Hong, the company’s chief executive, and Ji Yichao, its chief scientist, had been barred from traveling internationally following a meeting with China’s National Development and Reform Commission in Beijing. Both men can move within the country but cannot cross its borders. The restrictions were imposed after Chinese officials questioned them about how Manus restructured itself before the Meta transaction, and whether that process violated laws governing technology exports and outbound investment.
The move marked a significant escalation. What had begun in January as a formal regulatory review by China’s Ministry of Commerce had become, in the span of weeks, a direct personal restriction on the company’s founders.
For the governments, companies, investors, and engineers watching from both sides of the Pacific, the message was hard to miss: Beijing is prepared to use exit bans as a tool to enforce its claim over technology that was built on Chinese soil, even after a deal has closed and an entire team has moved to another country.
A startup built in China, sold in Singapore
Manus began as a product of a Beijing company called Butterfly Effect Technology, which was founded in 2022 and also operated a product called Monica.im. Over the following three years, Butterfly Effect built what it described as a general-purpose AI agent that could act on a user’s behalf across a wide range of tasks.
The company claimed its system outperformed OpenAI’s Deep Research agent on certain benchmarks. By the time it launched publicly in early 2025, it had attracted attention from major technology companies, including Microsoft, which began testing Manus in Windows 11 PCs in October of that year.
Manus reported that it reached $100 million in annualized recurring revenue within eight months of launch, with a revenue run rate that subsequently exceeded $125 million. In April 2025, the company raised a $75 million Series B round led by U.S. venture firm Benchmark. It was also backed by Tencent and HongShan Capital Group, formerly known as Sequoia’s China operation.
But the company’s Chinese roots were a problem. American lawmakers had been pushing to restrict U.S. investors from funding Chinese AI companies, and Benchmark’s participation drew scrutiny from members of Congress. Manus and its parent company responded by doing what a growing number of Chinese tech founders had tried to do: they relocated.
Through the middle of 2025, Manus closed its Wuhan and Beijing offices, deleted its Chinese social media accounts, let go of Chinese employees outside the core team, and moved that team to Singapore.
The company reincorporated there, replacing Chinese investors with foreign capital and restructuring its ownership so that, on paper, there were no continuing Chinese ownership interests. When Meta approached the company about an acquisition, Manus could present itself as a Singapore business. The deal was announced on December 29, 2025.
‘Singapore washing’ and the limits of corporate restructuring
In the months before and after the acquisition, the strategy Manus employed acquired a name in technology and legal circles: Singapore washing. The term describes the practice of Chinese founders building a product and a team in China, then reincorporating in Singapore to gain access to Western capital and to reduce the political and regulatory friction that comes with Chinese ownership.
Singapore has long served as a regional hub for multinational companies, and its legal and financial infrastructure makes it a credible home for technology businesses. For Chinese founders trying to raise money from U.S. investors or sell to U.S. acquirers, a Singapore incorporation offered distance from Beijing’s regulatory reach, at least in theory.
Matthias Hendrichs, a Singapore-based adviser to global AI firms, told CNBC that simple reincorporation is not sufficient. According to Hendrichs, the full team needs to relocate, the customer base must be transitioned, and early Chinese investors typically need to exit their positions before the restructuring is credible. Yuan Cao, a Beijing-based lawyer at the Yingke law firm, told CNBC that where a company builds its product matters more than where its holding company is registered. In cases like Manus, Cao said, developing technology in China before transferring assets to an offshore entity through restructuring is a red flag for regulators.
Beijing’s current position appears to reflect exactly that view. Chinese authorities are examining whether Manus transferred its core AI intellectual property out of China without the government approvals required under the country’s Regulations on Technology Import and Export Administration.
The investigation also covers potential violations of cross-border currency flow rules, tax accounting requirements, and overseas investment reporting obligations, according to reporting by financial news outlet Caixin and others.
Xin Sun, a senior lecturer in Chinese and East Asian business at King’s College London, told Fortune that Singapore washing is only effective for companies that fully cut off their operational ties to China. Manus, Sun said, relocated everything to Singapore in anticipation that its future market would be outside China. Whether that relocation was thorough enough to satisfy Chinese regulators is now the central question of the investigation.
A timeline of escalation
China’s scrutiny of the Manus deal unfolded in stages. When Meta announced the acquisition in late December 2025, the Chinese government initially said nothing. That silence ended on January 8, 2026, when the Ministry of Commerce issued a statement saying the deal would need to comply with Chinese laws on export controls, technology transfer, and outbound investment. The ministry’s statement was careful in its phrasing, reflecting what analysts at Geopolitechs described as a difficult balancing act between discouraging foreign acquisition of Chinese AI technology on one hand, and not frightening away domestic founders and talent on the other.
By January, the most likely outcome appeared to be a slow regulatory process that might delay integration but would not necessarily block the transaction. Meta had already begun folding Manus into its Superintelligence Labs division, with more than 100 Manus employees moving into Meta’s Singapore office in early March 2026. A Meta spokesperson told Reuters in an emailed statement that the transaction complied fully with applicable law and that the company anticipated an appropriate resolution to the inquiry.
The March meeting with the NDRC changed the situation. Following that questioning session, Xiao and Ji were told they could not leave China. The restrictions transformed the case from an administrative review into something more personal and more direct. With the co-founders physically separated from Meta’s global operations, integration of the company’s leadership is effectively stalled.
One source told the Financial Times that while extreme penalties for these specific violations are rare, an extreme outcome could involve trying to unwind the entire deal. The same source described such a scenario as messy given how far integration has already progressed.
A precedent-setting case in a fractured tech world
Legal analysts and investors are treating the Manus case as something more than a dispute between two companies. They see it as a test of how far China is willing to go in policing the movement of AI technology and the people who build it.
Wayne Shiong, managing partner of Argo Venture Partners, a Silicon Valley-based seed investor in AI, told CNBC that the path Manus took is one that people will not go down anymore. The exit bans have injected a new set of risks into a model that had become relatively common among Chinese founders trying to attract Western investment.
The Manus case has precedent in other cross-border technology disputes. Qualcomm’s attempt to acquire NXP Semiconductors collapsed in 2018 after Beijing withheld regulatory approval. Intel’s acquisition of Tower Semiconductor failed for similar reasons. In each of those cases, the obstacle was geopolitical friction rather than antitrust concern. What the Manus situation adds is the use of exit restrictions targeting individual executives, a more personal form of pressure that raises questions about travel risk for any Chinese founder or engineer involved in a foreign deal.
The broader context is a U.S.-China technology rivalry that has grown sharper over the past several years. The United States has used export controls to limit China’s access to advanced semiconductor chips, which are essential for training and running large AI models. Beijing has responded with its own restrictions, increased domestic subsidies for semiconductor production, and, as the Manus case illustrates, a more aggressive posture toward the outflow of AI talent and intellectual property.
According to analysts at the Center for Strategic and International Studies, the area where the United States holds its most significant advantage over China in AI is computing power. On other dimensions of the competitive landscape, including data, talent, algorithms, and energy, the gap is narrower. The Manus situation reflects Beijing’s awareness that talent mobility is one of the vectors through which it could lose ground.
What the case means for engineers and founders
For the generation of Chinese engineers driving innovation in AI, the Manus case has arrived as an uncomfortable signal. A cohort of founders and technologists who had grown accustomed to moving between ecosystems, raising money from U.S. investors, and pursuing acquisitions by Western companies now faces a more constrained landscape.
Shenzhen-based investors and founders told CNBC that the case has produced confusion and concern in Chinese technology circles. Some legal advisers are already counseling clients that the Singapore washing model carries risks that were not fully apparent before. Others are exploring alternative paths, including partnerships with Middle Eastern sovereign wealth funds or regional investors less exposed to U.S.-China political friction.
The personal dimension of the exit bans also raises questions that go beyond corporate strategy. Xiao Hong and Ji Yichao are individuals caught between two governments asserting competing claims over what they built. Their situation illustrates the human cost of a technology rivalry that has increasingly been prosecuted not just through tariffs and export regulations, but through restrictions on where people can go and what they can do.
Civil rights organizations and digital equity advocates have long argued that the governance of AI needs to include communities that are affected by the technology, not just the corporations and governments that control it. The Manus case adds another layer to that argument. As the most consequential AI deals are now shaped by the priorities of rival states, the question of who speaks for the public interest becomes more complicated and more important.
What comes next
The outcome of China’s review of the Meta-Manus deal remains uncertain. Beijing has not publicly stated what resolution it is seeking. Meta has said it expects an appropriate resolution without specifying what that would look like. The investigation could result in fines, required modifications to how the deal operates, forced disclosure of intellectual property, or, in the most extreme scenario, an order to unwind the transaction.
What is already clear is that the case has changed the calculations for Chinese AI founders, U.S. technology companies pursuing acquisitions in the sector, and investors on both sides. Singapore, which had positioned itself as a neutral hub for global AI businesses, now faces questions about whether its neutrality can be sustained in a world where the United States and China each claim jurisdiction over technology developed within their influence.
For Meta, the outcome matters beyond the balance sheet of a single acquisition. The company has committed over $115 billion in AI infrastructure spending in 2026. Its ability to execute that strategy depends, in part, on its capacity to attract and integrate talent from around the world. A prolonged standoff with Beijing, or a forced unwinding of a completed deal, would test that capacity in ways the company had not anticipated when it announced the Manus transaction three months ago.
Whether Xiao Hong and Ji Yichao will eventually be permitted to leave China and resume their work is a question only Beijing can answer. But the answer, when it comes, will carry consequences well beyond the two of them.
Disclosure statement
Insight News uses AI-assisted tools in some editorial production. Read our AI Policy & Transparency Statement at www.insightnews.com/ai-policy