China AI Startup Manus Raises More Than $500 Million After Meta Exit
The funding round, led by Boyu Capital and IDG Capital, gives the AI-agent developer new resources to compete in the U.S. and Chinese markets.

BEIJING — Chinese-founded artificial intelligence startup Manus has raised more than $500 million in a new funding round after resuming independent operations following the unwinding of Meta's planned acquisition of the company.
Butterfly Effect, the parent company of Manus, said Thursday that it had completed the funding round, which was co-led by Boyu Capital and IDG Capital. Existing investors including Tencent, Sequoia China and ZhenFund also participated, according to Reuters.
Manus develops general-purpose AI agents capable of carrying out tasks such as research and automation with limited human input.
The financing represents the company's first major funding round since its separation from Meta and provides fresh capital as Manus competes in the rapidly developing global market for AI agents.
Meta had previously agreed to acquire Manus in a deal worth more than $2 billion. However, Beijing ordered the acquisition to be unwound in April amid increased scrutiny of U.S. investment in Chinese startups developing advanced artificial intelligence technologies.
Manus said in August that it would resume operating as an independent company following the separation from Meta.
The company has continued to develop AI-agent products as competition intensifies among technology companies in the United States and China.
The Information reported that Manus' annualized revenue run rate had risen to around $500 million, compared with about $100 million when Meta acquired the company, according to The Information.
The latest funding gives Manus additional resources to expand its products and compete with major technology companies developing autonomous AI assistants.
The deal also highlights growing investor interest in AI-agent companies, which seek to move beyond conventional chatbots by enabling software to independently perform multi-step tasks for users.

