Meta to Buy AI Startup Manus for more than $2 Billion

Meta is buying Manus, an AI agent company, and the deal signals a clear shift in how Meta plans to use artificial intelligence. Meta is no longer focused only on AI that answers questions in conversation. Instead, it wants AI assistants that can actually do work for people.

This represents a change in strategy. Meta isn’t just building another product upgrade; it wants AI to become something people rely on every day, like a tool they truly need. Reuters and the AP both reported the deal, noting the price is believed to be in the billions of dollars.

The main ideas are simple to understand, but the more important message lies deeper. Manus is gaining momentum, it carries geopolitical complications, and it gives Meta a faster path to “AI that takes action” across consumer and business products.

The Price Tag

Meta and Manus have not disclosed financial terms, but estimates suggest the value is substantial. Some reports place Manus’s valuation between $2 and $3 billion, while another report cited by the AP put it above $2 billion.

That number matters because Manus is not simply a talent acquisition. The company sells a general-purpose AI agent that can perform tasks like research and coding. Manus has said demand for its agent has been unusually strong for such a young product. Manus reported more than $100 million in annual subscription revenue within about eight months of launch.

Meta is paying for speed. Building something people trust, rely on, and are willing to pay for takes time. Manus offers a ready-made starting point, an agent brand, a subscription engine, and a user base that already expects more than basic conversational AI.

A “Clean Break” From China

As Manus has grown, its origins have become more sensitive. The company was founded in China before moving its headquarters to Singapore. That history matters, especially when AI agents can access accounts, files, and active work.

Meta is pushing a clean break. After the acquisition, there will be no Chinese ownership interests, and Manus will stop operating in China. However, the company will continue to operate from Singapore.

Crucially, new Manus staff will not have access to historical customer data, and there will be restrictions on where models can be used. These controls aren’t just regulatory checkboxes; they are part of a broader plan to ensure Manus’s action-based AI can function within global data and legal limits, as Business Insider has reported.

For everyday users, this is the part of the story they never see, but it will determine whether AI agents gain mainstream adoption. People will only trust agents with real tasks if they believe data protections and safeguards are real.

Why Meta Needs Action, Not Chat

Meta has spent the past year pushing Meta AI deeper into Facebook, Instagram, and WhatsApp. The next step isn’t just making Meta AI smarter, it’s making it useful when people need to get something done, not just when they’re scrolling or asking questions.

That is where Manus fits in. Agents are supposed to do a job from the beginning to the end. This means they can do things like gather information, write code or take care of business workflows. If Meta integrates Manus’s capabilities into Meta AI, then the AI assistant can do more than just answer questions, like a chatbot.

This move also aligns with Meta’s broader AI roadmap. The company has shown it’s willing to spend heavily to accelerate progress, including its deal with Scale AI earlier in 2025. Manus serves a similar purpose: making AI more practical and effective in the real world.

The real test for Meta isn’t the announcement. It’s whether the company can deploy AI agents at scale without repeating past mistakes: users not trusting agents because they feel risky, and unresolved questions around data control and governance.

If Meta succeeds, 2026 might be remembered as the year Meta AI finally starts to actually work.