Meta's move is as unexpected as it is risky, revealing its concern about getting ahead in the race to achieve Artificial General Intelligence, something Manus has been boasting about since its founding in March 2025. What Meta's acquisition of Manus entails Although neither Meta nor Manus has officially confirmed the figure, the deal is believed to have closed for around $2 billion, an amount that aligns with the valuation the startup was seeking for its next funding round and quadruples the $500 million post-money valuation achieved just weeks after its launch, following a $75 million round led by Benchmark. In practice, Meta secures an asset that already generates over $100 million in annual recurring revenue and a total revenue rate exceeding $125 million, just eight months after its commercial launch. This would make Manus the fastest startup in the world to go from $0 to $100 million in ARR. In the competitive landscape, this move puts pressure on OpenAI, Google, and Anthropic in the specific area of ​​generalist autonomous agents, a segment where Manus had boasted of outperforming tools like OpenAI's Deep Research in complex research tasks. For Meta, incorporating an agent that already functions as an AI "execution layer" in real-world environments is a quick way to strengthen its generalist agent strategy against rivals who remain focused on the foundational model and chatbots, rather than end-to-end automation. The founders' Chinese origin is a source of unease in the US.

The deal comes amid added political noise: the founders' Chinese origins—they created the parent company Butterfly Effect in Beijing in 2022 before moving to Singapore—had raised concerns in Washington, to the point that Republican Senator John Cornyn publicly criticized US capital for funding "our greatest adversary in AI." To mitigate regulatory concerns, Meta told Nikkei Asia that, following the transaction, Manus will divest its Chinese ownership and suspend its services and operations in China, attempting to shield the acquisition from potential national security concerns in the United States. From a corporate perspective, the company will continue operating from Singapore, with a team of approximately 105 people spread across Singapore, Tokyo, and San Francisco, and plans to open an office in Paris, while maintaining its subscription service through its website and app (plans starting at €20 per month). Meta has indicated that Manus will continue to operate independently, while its agents will be progressively integrated into consumer and business products such as Facebook, Instagram, WhatsApp, and Meta AI itself, with the goal of scaling the solution to millions of businesses and billions of users. An Autonomous End-to-End Solution Part of the interest (and hype) surrounding Manus stems from how its technology works. The flagship product is an autonomous general-purpose AI agent designed to complete tasks and deliver final results, not just answer questions. It runs on a kind of virtual machine with its own internet access, persistent file system, and the ability to install software and create custom tools. The idea is that users can delegate end-to-end market research, programming, or data analysis tasks, while the agent plans, executes, and delivers production-ready work products, "like a virtual colleague with its own machine." Since its launch, Manus has processed over 147 billion tokens and created more than 80 million virtual computers, illustrating the volume of tasks it is already handling in the real world. The company boasts over 20% month-over-month growth since Manus version 1.5, supported by innovations such as Wide Research—designed for broad and structured research—and an approach that has popularized "context engineering" for AI agents, which, according to the company, has become an industry standard. Manus integrates with almost everything The key to Manus lies in its integrations: the agent can connect with Gmail, Notion, Stripe, HubSpot, Slack, Google Calendar, Google Drive, GitHub, Hugging Face, and other services using pre-built MCP connectors (an open standard designed to connect AI assistants with external tools), and also with internal systems through custom MCP servers. Based on this, the platform becomes a central orchestration layer for the user's digital environment, capable of reading data from different applications, executing actions across multiple platforms, and delivering results directly into existing workflows, eliminating the need for copying and pasting information. In addition to native connectors, Manus offers integration with Zapier to chain event-triggered automations and with Slack to send notifications, updates, and results directly to team channels. It also exposes its own API for integration into custom software. In practice, this enables use cases such as automated reporting, multi-app task creation, data synchronization between internal CRMs and external tools, and the generation of mobile apps from natural language—all triggered by a single prompt.