According to Business Insider, Meta announced on Monday it is buying the Singapore-based AI startup Manus. The Wall Street Journal reports the price tag was over $2 billion. Manus went viral in March for an AI agent that can do tasks like screen resumes and it relocated from China to Singapore in mid-2025. The startup claims to have processed over 147 trillion text tokens, has “millions” of users, and crossed $100 million in annual recurring revenue just eight months after launch. Meta plans to keep selling Manus as a standalone service while also integrating its tech into platforms like Facebook and Instagram.
Meta Buys a Business, Not Just Tech
Here’s the thing: Meta isn’t just buying fancy AI tech here. It’s buying an actual, functioning business with paying customers. That’s huge for them. Meta’s entire empire is built on giving stuff away for free and monetizing attention. Manus charges users up to $200 a month. So suddenly, Meta has a proven, high-margin software revenue stream it didn’t have to build from scratch. An analyst called it a “ready-made, high-margin software layer.” That’s the immediate win. Meta can point to this and say, “See? Our AI investments are paying off,” even while its internal “superintelligence” projects burn cash with little to show.
The Real Bet Is on AI Agents
But the bigger, riskier bet is on AI agents. Meta has totally failed to wow anyone with its own foundational models. It’s not OpenAI or Google. So Zuckerberg is pivoting. The theory, echoed by folks like Yuchen Jin, CEO of Hyperbolic, is that the real value isn’t in the base model—it’s in the smart software built on top. Manus primarily uses other companies’ models, like Anthropic’s Claude, and adds its own agent layer. Meta is basically admitting it lost the model war and is now trying to win the application war. It’s a smart hedge, but it also means Meta’s core AI destiny is still tied to its rivals’ technology. What happens if OpenAI or Google decides to cut off access or compete directly?
Distribution is Meta’s Only Card
So what does Meta bring to the table? Distribution. Billions of users across Facebook, Instagram, and WhatsApp. The plan seems to be: slap Manus’s “general-purpose agents” into every corner of those apps and hope it sticks. Zuckerberg already said Facebook is now a “broad discovery and entertainment space.” An AI agent that can plan your trip or manage your small business ads fits that shift. But, and this is a massive but, Meta has a terrible track record of integrating acquisitions in a way that users actually love. Remember WhatsApp? It’s basically the same app it was a decade ago. Instagram is just a storefront for ads and Reels. There’s a real risk Meta just smothers Manus’s agility and turns it into another clunky feature buried in a settings menu.
A $2B+ Shortcut With Big Risks
Look, this is a classic Zuckerberg move: see a gap, write a huge check to fill it. It gives Meta instant AI revenue and a story to tell investors. But it doesn’t solve the fundamental problem: Meta is not an AI innovator. It’s a distribution and monetization engine. Buying Manus is an expensive admission of that. The integration will be messy. The culture clash between a fast-moving startup and Meta’s giant bureaucracy will be real. And at the end of the day, Meta is still relying on other companies for the core AI magic. This isn’t a cure-all. It’s a very expensive, very desperate shortcut. Will it work? Probably not the way Zuckerberg is hoping. But for now, it gives him something new to talk about.
