On December 29, 2025, Meta said it will acquire Singapore-based AI agent startup Manus, which builds a general-purpose autonomous agent for businesses. Financial terms were not disclosed, and Meta plans to keep selling Manus as a standalone service while integrating it into Meta AI across its social and messaging apps.
This article aggregates reporting from 6 news sources. The TL;DR is AI-generated from original reporting. Race to AGI's analysis provides editorial context on implications for AGI development.
Meta buying Manus is one of the clearest signs yet that the frontier model race is morphing into an agent race. Manus isn’t just another chatbot; it’s a general-purpose AI agent already deployed as a “digital employee” that autonomously executes research, planning and workflow tasks for businesses, and it has reportedly handled massive token volumes and virtual machine spins in production. By absorbing Manus, Meta gains a mature agentic stack plus a team that has learned the hard lessons of real-world autonomy, rather than trying to bolt agents onto Llama from scratch. ([reuters.com](https://www.reuters.com/world/china/meta-acquire-chinese-startup-manus-boost-advanced-ai-features-2025-12-29/))
Strategically, this pushes Meta closer to OpenAI’s and Google’s visions of assistants that can operate end-to-end workflows with minimal supervision. If Meta successfully folds Manus into Meta AI and its enterprise offerings, it can turn its huge user base and infrastructure into a distribution and data advantage for agentic systems. That, in turn, could accelerate the feedback loop between model capability, tools, and real-world deployment—exactly the loop that tends to move the needle toward more general intelligence.
For competitors, the message is blunt: top-tier agent startups are now acquisition targets, not just API partners. Startups building autonomous systems will either need a very strong moat or a clear exit path, and cloud hyperscalers will feel growing pressure to own full-stack agents rather than leaving that layer to others.