What is Manus, and why did China block Meta’s $2 billion AI-agent deal?

The short version

Manus is a Singapore-based AI startup with Chinese roots. TechCrunch reported that it was founded by Chinese engineers and moved its headquarters from China to Singapore around mid-2025.

The company works in what the industry calls agentic AI: software meant to carry out more complete digital tasks for consumers and businesses, rather than simply produce text in response to a question. VentureBeat described Manus less as a traditional assistant and more as an execution engine—an agent designed to plan tasks, invoke tools and move workflows forward.

That positioning is what made Manus stand out. TechCrunch reported that the company drew Silicon Valley attention in spring 2025 after a demo video showed an AI agent screening job candidates, planning vacations and analyzing stock portfolios.

How Manus differs from a chatbot

A normal chatbot is usually built around conversation: you ask, it answers. Manus is part of a newer product category built around delegation: you give the system a goal, and it tries to break that goal into steps and act on them.

That distinction matters. If an AI system can reliably complete multi-step digital work, it becomes more than a search box or customer-service tool. It can become the interface through which users compare options, prepare documents, analyze information, fill out forms or move between apps. VentureBeat’s framing is useful here: Manus has positioned itself as an execution layer, not merely a chat interface.

Why Meta wanted Manus

Meta’s interest can be understood in three parts.

First, Meta has distribution. Fortune reported that the idea was to fold Manus technology into Meta products, including the Meta AI assistant used across Facebook, Instagram and WhatsApp. Business Today similarly reported that the deal was intended to accelerate autonomous AI agents across Instagram, WhatsApp and Facebook.

Second, Meta was not just buying another model. VentureBeat interpreted the deal as a sign that large technology platforms are competing not only on model quality, but on who controls the execution layer of AI-powered work. In other words, the strategic prize is not just having a smarter AI—it is owning the place where AI actually gets things done.

Third, Meta may have been buying speed. TechCrunch reported that Manus said in mid-December 2025 it had signed up millions of users and was generating more than $100 million in annual recurring revenue from monthly and yearly subscribers. Those are company-announced figures, not the same thing as independently audited financial statements, but they help explain why a large platform might prefer acquiring a fast-moving agent team instead of building and validating everything internally.

What a $2 billion-plus price was really buying

Public reports put the value of the deal in the low single-digit billions. Fortune reported that Meta had agreed to acquire Manus in a transaction valued at more than $2 billion. Business Today said official financial terms were not disclosed, while citing Wall Street Journal and Reuters estimates that placed the value between $2 billion and $3 billion.

That price was not just for a flashy demo. It was for a combination of three things: a product direction focused on AI task execution, market attention and user momentum built during 2025, and the possibility of plugging those agent capabilities into Meta’s existing social and messaging platforms.

Why China blocked the deal

This is the part that needs the most caution. The public reporting supports a narrower conclusion than some headlines may suggest.

What is clear is that TechCrunch reported in April 2026 that China’s NDRC had blocked Meta’s $2 billion acquisition of Manus after a months-long probe. TechCrunch also reported that Manus was founded by Chinese engineers and relocated its headquarters from China to Singapore around mid-2025.

So even though Manus was based in Singapore by the time of the Meta deal, the transaction involved a Chinese-founded AI startup and a major US technology platform. That was enough to bring the deal under Chinese regulatory scrutiny, according to the reporting.

What the available reporting does not establish is a precise public legal rationale or a single political explanation for the block. The safest reading is: Chinese regulators blocked the acquisition, the deal should not be described as cleanly completed, and the public information cited here is not enough to support a more specific claim about motive.

What it signals for AI competition

The Manus story is less about one startup and more about where AI products are heading. The first wave of consumer AI was dominated by chat windows. The next fight is over agents that can complete work.

VentureBeat argued that Meta’s agreement to acquire Manus was one of the clearest signs that large platforms are moving beyond competition over model quality and toward competition over the execution layer of AI-powered work.

For users, that could mean AI becomes less of a place where you ask questions and more of a tool embedded inside the apps where you already communicate, shop, organize, search or work. For Meta specifically, the appeal was connecting that capability to Meta AI, Facebook, Instagram and WhatsApp. But because TechCrunch later reported that China’s NDRC blocked the deal, that integration path now carries major uncertainty.

FAQ

Is Manus already part of Meta?

Not in a simple, settled sense. Reports in late 2025 said Meta had agreed to acquire or announced the acquisition of Manus, but TechCrunch reported in April 2026 that China’s NDRC had blocked the $2 billion deal.

Is Manus valuable because it built a better AI model?

The public reporting emphasizes Manus’s agent product and execution layer more than a single underlying model. Manus is described as an execution engine that can plan tasks, invoke tools and advance workflows, rather than merely answer isolated prompts.

Why would China be involved if Manus was based in Singapore?

TechCrunch reported that Manus was founded by Chinese engineers and moved its headquarters from China to Singapore around mid-2025. The same report said China’s NDRC blocked Meta’s acquisition after a months-long review. The reporting cited here does not provide enough detail to explain the regulator’s exact legal reasoning.

What is the main takeaway?

Manus matters because it represents a shift from AI that talks to AI that does. Meta’s interest shows how valuable the execution layer could become for major platforms; China’s intervention shows that AI agent deals can also become sensitive cross-border transactions.