Meta has agreed to acquire Singapore-based AI startup Manus, marking one of the company's most significant moves yet into commercial AI offerings. According to Bloomberg, the acquisition is part of Meta CEO Mark Zuckerberg's strategy to convert the company's massive investments in AI into tangible products and revenue streams.
Manus develops a general-purpose AI agent that it sells to small and medium-sized businesses through a subscription model, a structure that Investing.com reports has helped the company reach more than US$100 million in annual recurring revenue just eight months after launch, contributing to a total revenue run rate that exceeds US$125 million.
No financial terms were officially disclosed, but LatePost says Meta paid "several billion dollars," making it likely one of the largest acquisitions in the company's history behind deals for WhatsApp (acquired in 2014 for around US$22 billion) and Scale AI (acquired in June 2025 for around US$14.3 billion). For context, Meta acquired Instagram in April 2012 for US$1 billion. Manus was reportedly approaching a US$2 billion valuation in a pending financing round before the acquisition, reflecting strong investor confidence in its technology.
According to Reuters, Manus's AI agent distinguishes itself by operating as a digital employee capable of executing research, automation, and other tasks independently with minimal prompting. The company has supported this claim with free demonstrations on social platforms where it emphasized its performance exceeded that of other AI agents, such as OpenAI's DeepResearch.
The startup was founded by Chinese entrepreneurs and later moved its headquarters to Singapore. According to Sina Finance, Manus has processed more than 147 trillion tokens and created over 80 million virtual computing instances since launch, data that highlights its scalability and technical footprint. Founder Hong Xiao has said that after joining Meta, Manus will continue operating from Singapore and maintain its decision-making and product roadmap independently, even as the team integrates into Meta's broader AI ecosystem.
Article edited by Jerry Chen
