- Type
- General-purpose AI agent
- Developer
- Butterfly Effect (Singapore)
- Founded
- 2022 (China); HQ relocated to Singapore 2025
- Co-founders
- Xiao Hong, Ji Yichao
- Key event
- Meta acquisition blocked by China (April 2026)
- Related
- Agentic AI, Devin AI, multi-agent systems
- Type
- General-purpose AI agent
- Developer
- Butterfly Effect (Singapore)
- Founded
- 2022 (China); HQ relocated to Singapore 2025
- Co-founders
- Xiao Hong, Ji Yichao
- Key event
- Meta acquisition blocked by China (April 2026)
- Related
- Agentic AI, Devin AI, multi-agent systems
Manus is a general-purpose AI agent developed by the startup Butterfly Effect, capable of autonomously performing complex, multi-step tasks across the web — from research and data analysis to code execution and document creation. Launched in early 2025, Manus became one of the most prominent AI agent products globally and the centre of a major geopolitical controversy when China's regulators blocked Meta's US$2 billion acquisition of the company in April 2026.[1]
Origins and Development
Manus was founded in 2022 with roots in China by entrepreneurs Xiao Hong and Ji Yichao. The startup relocated its headquarters to Singapore in 2025, though its parent entity Butterfly Effect remained registered in China. Manus was positioned as "the general AI agent" — a system that could take a high-level user instruction and autonomously break it into sub-tasks, navigate web interfaces, execute code, and deliver completed work without step-by-step human guidance.[1][2]
The product generated significant excitement in the Chinese and global tech communities during 2025, with observers comparing it to Devin AI and other autonomous coding agents. Manus distinguished itself through its general-purpose design: rather than specialising in coding or research alone, it aimed to handle any task a knowledge worker might perform at a computer.
The Meta Acquisition
In December 2025, Meta announced its acquisition of Manus in a deal reported at approximately US$2 billion, making it the third-largest acquisition in Meta's history after WhatsApp and Scale AI. At the time, Manus CEO Xiao Hong stated: "We're excited about what the future holds."[1][2]
The acquisition appeared to represent Meta's major push into agentic AI, supplementing its Llama language model family with a proven autonomous agent platform. Meta integrated Manus staff and tools into its operations following the deal's close on 29 December 2025.
Chinese Regulatory Block
In March 2026, reports emerged that the Chinese government was reviewing the Meta–Manus deal. Co-founders Xiao Hong and Ji Yichao were required to appear before Chinese officials in Beijing and were subsequently prohibited from travelling abroad.[2]
In April 2026, China's National Development and Reform Commission (NDRC), through its foreign-investment security-review office, issued a formal decision prohibiting the acquisition and requiring the parties to withdraw the transaction. The ruling was made under measures in place since 2021 that review foreign investments for national security concerns when Chinese-origin technology and talent are involved.[1][3]
The NDRC's decision made clear that offshore incorporation — Manus's relocation to Singapore — does not shield a deal from Beijing's authority when the underlying technology and talent originated in China, a structure critics had termed "Singapore washing."[2]
Return to Independence
On 11 August 2026, Manus confirmed it would "soon return to operating as an independent company" as the unwinding of the Meta acquisition concluded. As part of the separation, data generated by certain users on or after 29 December 2025 (the acquisition closing date) was scheduled for deletion between 23 and 24 August 2026, Singapore time. Manus stated the deletion was driven by regulatory compliance, not a security incident, and that data created before the acquisition was unaffected.[1][4]
The unwinding process had been underway for months prior: Meta had cut off Manus staff from its internal data systems and barred Meta employees from using Manus tools as steps toward operational separation.[2]
Technology and Approach
Manus is designed around the paradigm of agentic AI — AI systems that can plan, execute, and adapt autonomously over extended workflows. Rather than responding to individual prompts, Manus maintains a goal state, decomposes objectives into actionable steps, interacts with web browsers and software tools, validates its own intermediate results, and iterates until the task is complete.
This approach aligns Manus with the broader industry trend toward autonomous agents, placing it alongside systems like Devin AI (software engineering), AutoGen (multi-agent frameworks), and CrewAI (agent orchestration).
Geopolitical Significance
The Manus case is widely regarded as a cautionary tale for AI startups caught between the United States and China. Beijing's intervention signalled that Chinese-grown AI companies cannot freely sell to US tech giants regardless of their corporate domicile, and that national security reviews will extend to AI talent and intellectual property.[3]
As AFP reported, Manus has become "an example of how rivalry between global powers can ensnare young entrepreneurs' ambitions," with analysts noting that China's move sent "a clear signal that attempting to shed Chinese identity through overseas relocation was no longer viable."[3]
>See Also
References
🇲🇾 Manus's Singapore base places it in Malaysia's immediate neighbourhood, and the agentic AI capabilities it represents are directly relevant to Malaysia's digital workforce ambitions. Malaysian developers and knowledge workers have been among the early adopters of AI agent tools in Southeast Asia, and the government's AI Untuk Rakyat programme — launched through AI Malaysia Berhad in August 2026 — explicitly aims to broaden citizen access to leading AI applications.
The Manus case also holds regulatory lessons for Malaysia. As Malaysia builds its own AI governance framework under AI Malaysia Berhad and the amended Personal Data Protection Act 2010, the Manus precedent illustrates the importance of clear cross-border data transfer rules and foreign investment review mechanisms for AI companies. Malaysia's ASEAN-aligned approach to AI governance, which emphasises interoperability with frameworks like the EU AI Act while preserving national sovereignty, is directly informed by cases like Manus.
Malaysian startups in the agentic AI space, supported by ecosystem builders like MRANTI and 500 Global, are watching the Manus independence closely as a potential catalyst for a more competitive Southeast Asian AI agent market.
References
- ↑[Singapore-based Manus AI to leave Meta as China blocks Meta's $2B acquisition](https://technode.global/2026/08/12/singapore-based-manus-ai-to-leave-meta-as-china-blocks-metas-2b-acquisition) — TechNode, 12 August 2026
- ↑[Manus is breaking from Meta as China forces the $2 billion AI deal apart](https://qz.com/manus-independent-meta-acquisition-china-unwind-081126) — Quartz, 11 August 2026
- ↑[Caught between great powers: the cautionary tale of Manus](https://www.france24.com/en/live-news/20260811-caught-between-great-powers-the-cautionary-tale-of-manus) — AFP/France 24, 11 August 2026
- ↑[AI firm Manus to resume 'independent' operations after China blocks Meta deal](https://www.channelnewsasia.com/world/manus-resume-operations-china-meta-deal-6312716) — Channel News Asia, 11 August 2026