Manus Raises $500 Million After Beijing Unwound Meta's $2 Billion Deal

A $2 billion acquisition got cancelled by a government, and the target just raised $500 million on its own. That sequence is worth reading closely, because it shows what happens to an AI company's value when its planned exit is forcibly closed.

On Thursday, Butterfly Effect, the parent company of AI agent startup Manus, said in a WeChat post that it has raised more than $500 million. TechCrunch reports it is the company's first funding round since Meta was forced to call off its $2 billion acquisition. The company did not disclose a valuation. Reporting on valuation below comes from earlier press coverage, not from the company, so treat it as unconfirmed.

What Actually Happened

According to TechCrunch, Boyu Capital and IDG Capital led the round. Existing shareholders Tencent, HSG (formerly Sequoia China), ZhenFund and others also participated. The Next Web, citing Chinese business outlet Yicai, adds that the announcement gave no valuation and did not say how the money will be used. Per TechCrunch, the company said it will keep hiring both at home and abroad.

The back story, assembled from TechCrunch, The Next Web and PYMNTS:

• Mid-2025: Manus relocated its staff to Singapore. TechCrunch says the startup had gone viral after a demo of its AI agent the year before.
• December 2025: Meta announced a deal to acquire Manus. PYMNTS puts the announcement at December 29, 2025, and says The Wall Street Journal reported Meta was paying more than $2 billion. TechCrunch reports the startup was said to be pulling in annual recurring revenue of more than $100 million at the time.
• January 2026: PYMNTS, citing CNBC, says Chinese regulators were investigating the purchase, with the probe focused on whether it complied with China's export control laws.
• April 2026: Chinese authorities ordered the deal unwound. The Next Web says a security review office run by the National Development and Reform Commission, China's state planner, banned the acquisition on April 27. PYMNTS quotes the NDRC as saying it would "prohibit foreign investment in the Manus project" and required the companies to "withdraw the acquisition transaction."
• June 2026: The Next Web reports Meta cut Manus off from its internal data systems, and that Manus's early investors moved to buy the company back at the $2 billion valuation.
• August and September 2026: Manus said it would resume independent operations. PYMNTS says a September 1 blog post confirmed it had formally resumed, with the founding team still leading the company.

TechCrunch adds that the company is reportedly considering a Hong Kong listing. That is a report, not an announcement, and the company has not confirmed it in anything we reviewed.

The Number Nobody Has Confirmed

The $500 million is the confirmed figure. The valuation is not. TechCrunch says Manus was said last month to be in talks to raise $500 million at a $4 billion valuation. The Next Web, citing Bloomberg, says Manus was close to raising that amount at about $4 billion, which it calls roughly double the $2 billion Meta had agreed to pay.

If that figure holds, the arithmetic is striking. The company would be worth about twice its acquisition price only months after the acquisition was struck down. We cannot verify it, because Manus did not respond to TechCrunch's questions about valuation and the announcement itself was silent on it. Still, the direction of the Bloomberg report matters more than the exact number. Investors appear willing to value the independent company at more than the strategic buyer offered.

That does not mean Meta's price was wrong. A buyer pays for control, integration and the risk it takes on. A minority investor in a growth round pays for upside. The two numbers measure different things, and a valuation reported in a funding negotiation is not a closing price.

It also helps to separate what the cap table says from what the market says. The investors named in the round include Boyu and IDG as leads and Tencent, HSG and ZhenFund as returning backers. The Next Web reports that when Meta cut Manus off from its data systems, early investors moved to buy the company back at the $2 billion level. Continuity among backers suggests the people with the most information about the business chose to stay in, rather than sell at the original price. That is an inference from the reporting, not a statement any investor has made.

What the Unwind Cost

The financial headline is the raise. The operational cost shows up elsewhere. Both The Next Web and PYMNTS report that Manus had to delete certain user data as part of the separation. The Next Web says the deleted data covered what some users had created on or after December 29, 2025, the date Meta bought the company. PYMNTS says Manus told users the deletion was needed to comply with regulatory requirements in certain parts of the world.

For a general-purpose agent product, user-created data is often the work product: research, files, tasks and drafts. If a customer built workflows on Manus after the acquisition was announced, some of that material was removed when the deal reversed. Manus's own note to users, per PYMNTS, said it was continuing product iteration and preparing new features. TechCrunch reports it recently launched Manus 2.0, described as a new architecture built around a new harness. It also introduced Cue, a stand-alone app that gives personal AI agents their own email addresses, phone numbers, digital wallets and computers, with payments limited to caps the user sets.

The product roadmap is moving. The data policy question is the part a customer should ask about directly. Nothing in the reports says Manus mishandled data. They say it deleted some under regulatory pressure. For a buyer, that distinction does not change the practical result: the vendor's ownership changed twice in one year and the data followed.

Why the Regulator Mattered More Than the Buyer

The NDRC's language, as quoted by PYMNTS, is blunt: prohibit foreign investment in the project, withdraw the transaction. TechCrunch frames the motive as intensifying worries in China over losing AI talent and researchers to the West. Manus had moved to Singapore before the deal. The reversal suggests that relocating a company did not remove it from the reach of Chinese regulators.

The business consequence is direct. A planned exit assumed a buyer, a price and a close. In this case the buyer was willing and the price was set, and the exit still failed because a third party held a veto. The company that emerged is independent and newly funded, but it spent roughly nine months in limbo between announcement and the end of the deal.

For founders, that is a data point on exit risk for any AI company with ties to China, whatever its current address. For customers, it is a data point on vendor stability.

Where Manus Sits in the Market

TechCrunch places Manus alongside companies such as Cursor, Lovable and Replit: AI products and agents that let users build apps and websites, create designs and presentations, generate video, and more. The company offers a chatbot and vibe-coding tools. That is a crowded category, and a $500 million round puts Manus in a position to compete on hiring and product speed.

We would not read the round as proof of product-market fit beyond what the company has disclosed. The only revenue figure in the reporting is TechCrunch's note that Manus was said to have annual recurring revenue above $100 million at the time of the Meta announcement. That is a past figure from a different moment, and no current revenue number has been published in the reports we reviewed.

What Builders Should Take From It

• Treat acquisition as a risk, not a plan. If your exit depends on a strategic buyer, map every regulator that could block it, including ones outside the buyer's home country.
• Ask agent vendors where your data lives and what happens in a deal. Manus deleted some user data tied to the acquisition period. Get deletion, export and change-of-control terms in writing.
• Keep your work portable. If an agent product holds your files and workflows, export them on a schedule and keep a second tool that can read them.
• Separate confirmed from reported. The $500 million, the investors and the April ban are confirmed in the reporting. The $4 billion valuation and the Hong Kong IPO are reports. Do not put either in a board deck as fact.
• Watch the agent category for funding signals. A large round for a general-purpose agent company shows investors still back the category after a deal collapse. It does not show which product wins.

Developer312 covers the AI business signals builders actually need to act on. Get the weekday briefing at developer312.com.