Meta Unwinds $2B Manus Deal After Beijing Orders Reversal
Meta is dismantling its $2 billion acquisition of Manus after Beijing ordered the deal reversed. Here's what happened and what it means for AI dealmaking.
LUMIEN3 min read
Meta is moving to dismantle its $2 billion acquisition of Manus, the AI agent platform, after Beijing reportedly ordered the deal reversed, according to TechCrunch. The directive from Chinese authorities puts Meta in the unusual position of having to unwind a deal it had already agreed to, and it places Manus back in limbo just as AI agent tools have become one of the hottest investment categories in tech.
What happened
Meta agreed to acquire Manus, an AI agent platform, for $2 billion. After the deal was struck, Beijing stepped in and ordered it reversed. Meta is now reportedly working to dismantle the acquisition in response to that demand.
Manus attracted significant attention earlier in 2025 after going viral for its autonomous task-completion capabilities. The platform was built by a team with Chinese origins, which appears to be central to why Chinese regulators took an interest in the transaction.
Why it matters
A government ordering a buyer to unwind a completed acquisition is not a common event. It puts a spotlight on a risk that many companies doing cross-border AI deals have not fully priced in: regulatory reversal, not just regulatory delay.
For anyone watching the AI M&A market, this case raises three specific concerns:
- Deal certainty: If Beijing can order a $2 billion deal reversed after signing, acquirers will need to factor Chinese regulatory approval into timelines and structures for any target with Chinese founders, investors, or infrastructure.
- AI as a strategic asset: Chinese authorities treating an AI agent platform as something worth intervening over tells you how seriously governments now view AI capabilities as matters of national interest.
- Meta’s AI strategy: Meta has been aggressive about building and buying AI talent. Losing Manus, and the cost and distraction of unwinding the deal, is a real setback regardless of how the company frames it publicly.
The broader signal here is that the US-China technology divide is not just about chips and semiconductors anymore. AI software, particularly agentic AI, is now firmly in the same category.
Our take
From where we sit, this deal is a useful reminder that “closed” does not mean “safe” when one party has roots in a jurisdiction that views the technology as strategic. Due diligence now has to include a serious look at where a target’s founders are from, where its data lives, and whether any foreign government could plausibly claim an interest in blocking or reversing a sale.
That is not a hypothetical risk anymore. It happened here, at $2 billion, to one of the largest technology companies in the world. Smaller acquirers with fewer legal resources would be in a far worse position navigating the same situation.
We are also watching what this means for Manus itself. Being at the center of a reversed geopolitical acquisition is not a neutral event for a product’s roadmap, its team’s morale, or its ability to sign enterprise customers who now have every reason to ask hard questions about stability and ownership.
What to do about it
If your business is evaluating AI tools or vendors, particularly newer platforms that went from obscure to famous quickly in 2025, it is worth asking a few direct questions before you build a dependency:
- Where is the company incorporated, and where are its founders and key staff based?
- Who are its investors, and do any of them operate under jurisdictions that could impose restrictions on the product?
- What happens to your data and workflows if ownership changes or the product is forced to restructure?
None of this means you should avoid newer AI tools. It means you should know what you are building on before you build on it.