- Manas AI founders seek to raise $1 billion to buy back Meta after China orders Meta to back out of more than $2 billion in acquisitions
- The startup could resurface as a China-related joint venture before pursuing a possible IPO in Hong Kong
- The case is shaping up to be a landmark warning against U.S. AI deals involving technology of Chinese origin, and a human story about what happens when geopolitics throws founders into the crossfire.
The two founders of one of the fastest-growing companies in AI have been banned from leaving China while regulators investigate the sale of their startup to one of the world’s most powerful technology companies. The details, buried in the regulatory review machinery, tell us more about where this story is headed than any valuation.
China has blocked Meta Platforms’ planned acquisition of Manus AI worth more than $2 billion and ordered the deal to be dissolved. Now, Manas co-founders Xiao Hong, Ji Yichao and Zhang Tao are considering a funding round of up to $1 billion to take back control of the company they founded, Bloomberg reports, citing people familiar with the matter. The founders may also contribute their own funds to fill any remaining gaps, and the buyback is expected to be at a valuation that at least matches what Meta originally paid.
a broken deal
Manas was founded in China and moved its headquarters and core operations to Singapore in 2025. As TFN reported at the time, the move immediately drew scrutiny from Chinese regulators over whether it amounted to an attempt to move strategically sensitive technology out of the reach of the Chinese government. When Manus first launched in March 2025, it was hailed as the world’s first general-purpose AI agent, and within days millions of people were on waiting lists. It had already been called China’s answer to deep-seeking, but this label makes Beijing’s decision to block the Meta deal seem even more deliberate.
The company has developed systems that can autonomously process complex digital tasks and workflows with limited human input, making it one of the companies defining the emerging agent AI market. The rise was unusually steep. The startup reportedly surpassed $100 million in annual recurring revenue within nine months, a trajectory that garnered serious attention from investors and, ultimately, the interest of Meta.
Meta acquired Manus in December to deepen advanced AI integration across its ecosystem. However, China’s National Development and Reform Commission has launched an investigation into whether the deal violates China’s investment rules, particularly regarding the transfer of strategically important AI technology to the US buyer. The two co-founders of Manus have reportedly been banned from leaving China during the investigation, details that turn what appeared to be a corporate dispute into something much more serious.
Meta has not publicly commented on the termination of the agreement or what Manas has committed to its AI roadmap. A Meta spokesperson said in a brief statement that the transaction was “in full compliance with applicable law” and the company looked forward to “an appropriate resolution to the investigation.” However, this acquisition was no accident. Agenttic AI (systems that can plan, reason, and execute tasks autonomously) is increasingly seen as the next frontier in enterprise software, and Manus has been one of the most trusted players in the field. What the meta loses here is more than just a target. This is a stepping stone into a category where rivals, including OpenAI, are competing to dominate.
The more difficult problems are technical
Financing share buybacks is only part of the challenge. Much of Manas’ technology is already integrated into Meta’s systems, making complete separation technically complex and expensive. The proposed funding would have to cover not only the buyback, but also the cost of dismantling the startup’s infrastructure and rebuilding it as a truly independent business. The process has been described as arduous and unpredictable by engineers and legal advisers in similar situations.
It’s unclear whether raising $1 billion is achievable. The company has been repurchased under pressure from regulators, its technology has been partially integrated into other companies’ systems, and the geopolitical environment that killed meta trading remains the same. Investors need to weigh the upside. Manas is expected to generate revenue of about $1 billion in 2026, according to people familiar with the company’s finances. Compare that to a risk profile that is truly unusual even by late-stage AI investing standards.
What happens next and what it suggests
Once the buyback is complete, Manas is expected to resurface through a Chinese joint venture structure backed by new investors, with an IPO in Hong Kong rather than a U.S. listing likely its next destination. The choice of venue itself is important. Hong Kong sits at the intersection of Chinese regulatory jurisdiction and international capital markets, and listing in the U.S. poses obvious complications for companies with technology that China currently declares strategically sensitive.
The broader implications extend far beyond Manus. This is now one of the clearest cases on record in which China has actively blocked the transfer of AI technology to a US acquired company, directly restricting the personal freedoms of the founders involved. For venture capitalists backing Chinese AI startups, for U.S. technology companies evaluating acquisitions in the space, and for startups themselves navigating dual jurisdiction, the questions this case raises are no longer hypothetical.
The next time a fast-growing AI company relocates from China to Singapore and raises money from Western investors, its founders, their lawyers and backers will all be thinking about what happened to those who couldn’t leave.
