Meta’s plan to acquire the AI startup Manus for $2 billion has hit a major regulatory hurdle in China. Although Manus is now headquartered in Singapore, the company was originally founded in China, and regulators are concerned that its relocation may have bypassed critical export-control laws. The Chinese government is currently reviewing whether Manus needed an official export license to move its technology and operations abroad. If investigators find that the company violated these laws, the founders could face criminal liability and the entire deal could be blocked.
The scrutiny highlights a growing trend known as “Singapore washing,” where Chinese startups relocate to the city-state to avoid geopolitical tensions and make themselves more attractive to Western buyers like Meta. Meta is interested in Manus because of its expertise in “general-purpose AI agents”—software that can perform complex actions rather than just generating text. Manus has already processed over 147 trillion tokens and created 80 million virtual computers, making it a valuable asset for Meta’s goal of moving AI from simple conversation to direct action.
Under the current terms of the deal, Manus is expected to continue operating as a separate brand based in Singapore. CEO Xiao Hong stated that joining Meta would provide a more sustainable foundation for the company’s growth without changing its core decision-making process. However, if the Chinese investigation finds that the technology was exported illegally, Meta may find it difficult to integrate Manus’ intellectual property into its global portfolio. This could stall Meta’s efforts to improve its AI assistant and push ahead of competitors in the agentic AI space.
So far, there has been no public resistance to the deal from United States regulators, as the acquisition aligns with efforts to bring advanced AI talent into the American tech ecosystem. The outcome now rests on the Chinese review of Manus’ original export permits. If the deal proceeds, it would mark one of Meta’s most significant investments in AI hardware and agent-based software to date.
