On December 30th, Beijing time, Manus officially announced that "Manus will soon be joining Meta."
For Manus, a startup, the acquisition by Meta is a recognition of its work in the field of general AI agents. For Meta, this is the third largest acquisition in its history, with a price tag of several billion dollars, second only to the $19 billion acquisition of WhatsApp in December 2014 and the $14 billion acquisition of Scale.ai in June 2025.
According to official statistics from Manus in early December, since its launch, Manus has processed over 147 trillion tokens and created over 80 million virtual computers. In this collaboration with Meta, Manus stated that it will further solidify Manus's strategic position in the AI application layer—deploying artificial intelligence... The capabilities translate into scalable, reliable systems that execute user-assigned tasks end-to-end in real-world use cases. To ensure this acquisition does not disrupt normal user experience, Manus will continue to provide products and subscription services through its app and website, and the company will continue to operate in Singapore.
Manus CEO Xiao Hong stated, "Partnering with Meta allows us to grow on a stronger and more sustainable foundation without altering Manus's operating methods and decision-making mechanisms. We are very excited about the prospects of the collaboration between Meta and Manus." Following the acquisition, Butterfly Effect will continue to operate independently, and founder Xiao Hong will become Meta's Vice President.
Liu Yuan, a partner at ZhenFund and an angel investor in the Butterfly Effect, said that the acquisition took only a little over ten days and will be a great inspiration to young entrepreneurs in the AI era.
On the other hand, the acquisition represents another gamble by Meta, which has been aggressively pursuing growth this year. Before Manus achieved widespread popularity, ByteDance offered approximately $30 million to acquire it from Xiao Hong in 2024. In the past two years, the foundational large-scale model technology has continuously iterated, the AI application ecosystem has been enriched, and Manus has achieved widespread dissemination thanks to its product performance. According to reports, before Meta expressed its acquisition intentions, Butterfly Effect was attempting to complete a new round of financing with over $100 million in ARR (Annual Recurring Revenue).
From tens of millions of dollars to billions of dollars, the soaring valuations reflect both the growth of Chinese AI entrepreneurs and the anxiety of US tech giants like Meta. In 2025, Meta attempted a high-profile strategy across model iteration, computing infrastructure, organizational restructuring, and mergers and acquisitions to catch up with OpenAI and Google, but with limited success. In early December, Morgan Stanley... We have lowered our target price for Meta from $820 to $750.
In terms of technical roadmap, the entire large-scale model race has reached a bottleneck in the growth of basic large-scale models, transitioning from a phase of "unlimited data" to one of "limited data." Sebastian Borgeaud, one of the leads in Gemini 3 pre-training, previously stated that while scale remains important, the weight of architectural and data innovation has significantly increased, even becoming more critical. At this juncture, Manus, with its substantial subscriber base and engineering capabilities, has naturally come into Meta's view.
By acquiring Manus, Meta can quickly address its shortcomings in productization and commercialization in the field of general-purpose intelligent agents and accelerate the construction of its internal ecosystem. However, this acquisition cannot fundamentally solve Meta's persistent problems in areas such as top-tier model performance, computing infrastructure, and even corporate issues. The multi-billion dollar acquisition is merely another move by Meta driven by anxiety, rather than a "magic bullet."
Cheng Weizhong, founder and CEO of Zhongke Shenzhi, told CBN reporters that this acquisition shows that industry competition is gradually shifting from base models to application-side applications.
Zhang Fa'en, CTO of AInnovation, told reporters that the acquisition of Manus is significant for Meta primarily because it greatly enhances the capabilities of Meta's general-purpose agent platform. He stated that Meta's choice was wise because the base model and agent can be trained separately. Especially in the base model domain, open-source models such as MiniMax M2.1 and GLM 4.7 currently offer strong performance. Manus brings the capabilities of a general-purpose agent to Meta, offering significant potential and helping Meta generate a ripple effect on platforms like Facebook.
In a recent interview, Xiao Hong mentioned the relationship with tech giants. He stated that Manus's survival strategy was to cooperate and coexist with these giants, acting as the "best experience integrator." Competition in the underlying model is fierce, and no single company can sustainably monopolize all capabilities. As the application layer, Manus can flexibly integrate the best models from various companies, theoretically providing users with a more refined experience than any single provider. This is similar to the relationship between mobile phone manufacturers and chip manufacturers: although they don't manufacture chips (models), their deep understanding of user needs and massive usage allow them to optimize models, creating a win-win situation. However, this acquisition by Meta may alter this positive cooperative relationship with the giants' foundational models.
(Article source: CBN)
