Manus seeks a $4bn valuation months after Beijing undid its sale to Meta

Manus, the AI agent developer whose sale to Meta was blocked by Beijing in the spring, is in talks to raise about $500mn at a valuation of roughly $4bn, according to a Wall Street Journal report relayed by TechCrunch on Friday, September 18. Prospective new investors named in the reports include IDG Capital, Boyu Capital and the battery maker CATL, while existing backers Tencent, HSG and ZhenFund are also said to be considering taking part. The company is also reported to be weighing a corporate restructuring that could prepare it for a listing in Hong Kong. Manus did not immediately comment.

The number matters less for its size than for what it implies. Only weeks ago a group of former investors bought the company back from Meta at about $2bn. If the round closes at the reported price, the same asset will have doubled in value in about five weeks, without any change in its product, simply by moving from a contested American owner to a cap table that Beijing can live with. That is a striking lesson in how geography, rather than technology, now sets the price of an AI startup with Chinese roots.

From viral demo to regulatory casualty

Manus was founded in 2022 by Xiao Hong, Yichao Ji and Tao Zhang, under a parent company, Butterfly Effect, that was established in Beijing. Its general purpose AI agent drew wide attention after a demonstration in 2025, and in the middle of that year the company moved its headquarters and most of its China-based staff to Singapore. In December 2025 Meta announced an acquisition valued at roughly $2bn to $3bn, at a point when Manus was reported to have passed $100mn in annual recurring revenue.

The deal never settled. China said in January that it would examine whether the transaction complied with its laws, and on April 27 the National Development and Reform Commission announced that it had decided "to prohibit foreign investment in the Manus project in accordance with laws and regulations" and had required the parties to withdraw from the transaction. The planner gave no reasons. Reuters reported that the chief executive and the chief scientist were under exit bans preventing them from leaving mainland China. Meta said at the time that the transaction "complied fully with applicable law".

What followed was an unusual reverse acquisition. According to Caixin, Tencent, ZhenFund and HSG were among the former investors who bought Manus back from Meta for $2bn. Benchmark, the American venture firm that had been the largest shareholder before the Meta deal, did not take part, and Tencent acquired its former stake to become the largest shareholder. In August Manus told users to export their data, saying that material generated after the acquisition had to be deleted to comply with "regulatory requirements in specific jurisdictions". In September the company said it had resumed independent operations under its founding team.

Why the price can double so quickly

The $2bn buyback was not a market price in any meaningful sense. It was a settlement struck under regulatory compulsion, with a motivated seller in Meta and a limited set of buyers acceptable to Beijing. Such transactions tend to anchor at the original deal value because that is the easiest number for all parties to justify, not because it reflects what the business is worth.

A fresh primary round is a different exercise. New money is being asked to price the company as an ongoing concern, and the presence of IDG, Boyu and CATL suggests that well connected Chinese capital sees an opportunity in an asset that American buyers can no longer touch. For the insiders who bought at $2bn, a $4bn round would crystallise a paper gain almost immediately and validate their decision to step in.

There is also a strategic logic. A group of domestic investors that includes Tencent and an industrial champion such as CATL gives Manus a shareholder base that aligns with the direction Beijing signalled when it vetoed the Meta deal. A Hong Kong listing, if it materialises, would complete that alignment by keeping the eventual exit inside a market under Chinese jurisdiction.

The business underneath

The harder question is whether the underlying company deserves the price. According to the Journal's reporting as summarised by Channel Insider, Manus does not train its own foundation models from scratch but builds on open weight models and post-training techniques. That is a capital light approach, which helps margins and speed, but it also means the core intelligence is rented. Its products, a chatbot, vibe coding tools for apps and websites and tools for design, presentations and video, sit in categories where OpenAI, Lovable and Replit already compete hard.

Bloomberg has also pointed out that Moonshot's Kimi and Anthropic's Claude are moving into general desktop tasks, which puts pressure on pricing for agent products. An application layer company without its own models faces the risk that the model providers absorb its features. Manus's answer has been speed of execution and distribution, which earned it the Meta offer in the first place. Whether that edge survives a year of disruption, a forced data deletion and a change of ownership is something the new investors will have to judge from the inside.

The disruption itself carries a cost. Users who were told to back up their data and saw content deleted for regulatory reasons may not all return. Enterprise customers, in particular, tend to prize continuity, and a vendor whose ownership was decided by a state planner is a harder sell in procurement meetings outside China.

A template for cross-border deals

The Manus episode has become a reference case for founders who build in China and then relocate abroad. Moving staff and headquarters to Singapore did not place the company beyond Beijing's reach, and the April decision showed that the origin of the founders and the technology can outweigh the domicile of the holding company. For American acquirers, the lesson is that a signed and announced deal can be reversed months later, with integration work and senior hires stranded.

For investors in Chinese founded AI companies, the picture is more nuanced. Benchmark's absence from the buyback, after scrutiny in Washington of its original investment, suggests that the shareholder register of such companies is being sorted along national lines. Domestic capital is picking up the stakes, and domestic exits are being prepared. That narrows the pool of potential acquirers and listing venues, which in normal circumstances would weigh on valuation. The reported $4bn price implies that investors either expect the Hong Kong route to deliver strong demand or believe that the business has grown enough since December to offset the narrower exit options.

Our view

The proposed round is less a vote on Manus's technology than a pricing of political certainty. Its new and returning backers are paying for an asset whose ownership question has been settled in Beijing's favour, and they are betting that a Hong Kong listing will reward that certainty. That may prove right, but founders elsewhere should take the narrower lesson: relocating a holding company does not change where a regulator believes a technology comes from. Western investors should treat Chinese rooted AI companies as having one realistic family of exits, and price them accordingly. What to watch: whether the round closes at the reported terms, which investors commit in writing, how much revenue Manus can show after the data deletion, and whether the restructuring for Hong Kong begins. A rapid doubling of value after a forced sale is a good story for insiders. It becomes a good investment only if paying customers stayed through the turbulence.

Sources

• TechCrunch: Manus seeks $4B valuation in new $500M fundraise as it resumes independent ops
• Channel Insider: Manus eyes $4B valuation while exploring Hong Kong IPO
• Caixin Global: Manus cuts ties with Meta as Tencent emerges as top backer
• TechCrunch: China vetoes Meta's $2B Manus deal after months-long probe

Filed under Startups. Written by InvestTimes Editorial under our editorial standards. Commentary and analysis for information only, not investment advice.