Manus Funding Round and Valuation
Manus, a Chinese‑founded artificial‑intelligence startup that builds generative AI agents, is seeking a valuation of about $4 billion in a planned $500 million financing round. This is the company’s first external capital raise since it completed an operational separation from Meta Platforms in May, when the two entities halted data sharing and ended their joint development arrangement.
The proposed funding could roughly double Manus’s current valuation, potentially making it China’s most valuable AI‑agent startup and positioning it for a future listing on the Hong Kong Stock Exchange. The identities of the prospective new investors have not been disclosed, and Bloomberg notes that discussions are still at an early stage with terms subject to change.
Manus’s existing backers are Tencent Holdings, HSG, and ZhenFund. Representatives for Manus, Tencent, HSG and ZhenFund did not respond to comment requests.
The financing initiative follows a regulatory intervention by Beijing that forced Manus and Meta to unwind their acquisition deal. Meta had agreed to acquire Manus after the startup surpassed $100 million in annualized revenue, but Chinese regulators blocked the transaction over concerns that the technology could be transferred to a geopolitical rival. After the forced unwind, Manus announced in early September that it had resumed independent operations, with its founding team continuing to lead product development for global users.
A $4 billion valuation would strengthen Manus’s competitive standing among China’s AI‑agent developers and could pave the way toward a Hong Kong listing, according to Bloomberg. The company’s rival, Evoken, which develops the AI design agent Lovart, is also raising capital at a reported $3 billion valuation.
Manus still faces competition from broader foundation‑model providers. Bloomberg highlights that products such as Moonshot’s Kimi and Anthropic’s Claude are increasingly capable of handling general desktop tasks, whereas Manus does not train its own base models from scratch.