AI dealmaking isn’t slowing down — and it’s Meta’s turn again.
The social media giant is buying Manus, a Singapore-based artificial intelligence startup, the companies announced on Monday.
Manus went viral in March when it previewed an AI agent that could autonomously perform tasks like screening résumés and stock analysis. Manus was created in China but relocated to Singapore in mid-2025. Meta paid more than $2 billion for Manus, the Wall Street Journal reported.
The deal is the latest in a flurry of red-hot AI investment and acquisitions this year, which includes Meta’s $14 billion investment in Scale AI in June. From providing an instant AI revenue source to giving it a leg up in AI agents, here’s why buying Manus could give Meta a much-needed boost in the AI race.
1. It’s an instant revenue generator
Manus said in December that it had processed more than 147 trillion tokens of text and said its users were in the “millions.” It also claimed to have crossed $100 million in annual recurring revenue, achieving both milestones eight months after launch.
Those numbers tell us Meta is getting a startup with a built-in audience of paying users. Meta’s business model to date has largely revolved around building free products and making money from collecting user data and targeted advertising. Manus offers a free tier for basic tasks, but charges users up to $200 a month for its pro tier.
“The purchase gives Meta a functioning business with paying customers, meaningful revenue and infrastructure already proven at scale,” said Murthy Grandhi, company profiles analyst at research firm GlobalData, in a note.
In its announcement, Meta said it plans to continue selling the Manus service separately while also integrating Manus’s technology into its existing platforms, which include Facebook, Instagram, and WhatsApp. Meta did not elaborate on which ones or how it might do so. Meta has poured billions into building up its internal AI teams and developing what is termed “superintelligence,” with so far little in terms of returns. Manus could be a way for Meta to start making money directly from AI while it continues to build out its internal efforts.
2. Manus is a big bet on agents
Meta has struggled to wow consumers and developers as much as OpenAI and Google when it comes to raw model power. However, as these models become increasingly commoditized, there is a growing need to show AI can actually be useful. One such way is AI agents, a type of software that can proactively make decisions and take actions, such as creating a marketing campaign or monitoring and fixing bugs in apps.
Buying Manus could prove a smart bet on the idea that the real value will lie in the programs that sit on top of the models. Manus primarily uses other companies’ AI models, like Anthropic’s Claude, as building blocks and layers its own software on top.
“People keep assuming a small update from OpenAI or Google will wipe out a lot of AI startups,” wrote Yuchen Jin, CEO of the AI startup Hyperbolic, in an X post about Meta’s Manus deal. “But in reality, the AI application layer should be where most of the opportunity is.”
A Meta spokesperson did not immediately respond to a question about which models Manus would support following the acquisition.
3. Meta can use its distribution advantage
One of Meta’s strengths is that its platforms are used by billions of people, which, like Google, gives it a distribution advantage. Its challenge is to find ways to keep them coming back.
Unlike Google with Gemini 3, Meta has yet to have a buzzy AI breakthrough moment with its own in-house models. Combining Manus’s “general-purpose agents” with Meta’s distribution channels gives the social media company another shot at that, particularly as CEO Mark Zuckerberg has acknowledged that Facebook has shifted from being a place for friends to view each other’s content to a “broad discovery and entertainment space.”
“Manus offers a ready-made, high-margin software layer that can be sold directly and integrated across Meta’s consumer and enterprise products,” said GlobalData’s Grandhi.
