Butterfly Effect says Manus raised over $500M. Valuation stays undisclosed

What did the 8 October reports say happened?

Kate Park’s TechCrunch story, published 8 October 2026 at 6:20 a.m. PDT, says Chinese AI lab Manus’s parent, Butterfly Effect, “said on Thursday in a WeChat post that it has raised more than $500 million,” the first funding round since Meta “was forced to call off its $2 billion acquisition.” Anniek Bao’s CNBC report the same day, timestamped 2:32 a.m. EDT and updated at 2:59 a.m. EDT, uses the same floor and the same parent-company attribution. Maria Deutscher’s SiliconANGLE note, updated 16:25 EDT, also says more than $500 million.

TechCrunch and CNBC both say Boyu Capital and IDG Capital led the round, with follow-on from existing shareholders Tencent, HSG and ZhenFund. TechCrunch adds “and others” and that the company said it will continue hiring at home and abroad. SiliconANGLE’s lead sentence names Boyu and Tencent and “several others,” omitting IDG as a co-lead. We are using TechCrunch and CNBC for the lead pair.

This is a financing story about an agent lab, not a new model card. For what “agent” means in product copy, see our guide to AI agents. For how to keep a vendor number in its column, see how to read an AI company announcement.

What is on Manus’s own English pages?

We opened manus.im/blog/manus-resumes-independent-operations and manus.im/blog/introducing-manus-2-0 on 9 October. Neither page announces a funding round, names Boyu or IDG, or states a valuation. The independence post is undated on the copy we fetched. It says Manus “has formally resumed independent operations,” the founding team will continue to lead, and some users had to back up and restore data. It points to a restoration portal and says there is no deadline to restore.

CNBC writes that “earlier this month” Manus said it had resumed independent operations. TechCrunch dates that restart to August and adds that Manus “said it was required to delete some user data as part of the split.” The English independence post we opened talks about backup and restore. It does not use the word delete. We are keeping TechCrunch’s delete line in TechCrunch’s column.

The 2.0 post is also undated on the copy we fetched. SiliconANGLE says Manus 2.0 “rolled out a few days before today’s funding announcement.” That timing is SiliconANGLE’s. The 2.0 page itself does not date the launch.

How do the wires describe the Meta unwind?

The 8 October stories treat the raise as the first round after a blocked Meta deal. TechCrunch’s recap: Manus went viral on an agent demo last year, moved staff to Singapore in mid-2025, then announced a $2 billion Meta acquisition that December. In April, “Chinese authorities ordered the AI startup to unwind the deal.” CNBC names the National Development and Reform Commission and quotes it as deciding to “prohibit foreign investment in the Manus project.” Those are CNBC’s words for the regulator, not a document we opened.

TechCrunch says Manus “was said to be pulling in annual recurring revenue of over $100 million” at the time of the Meta deal. That is a reported figure, not a number on the English blogs we opened. The same story says Manus “is also reported to be considering going public in Hong Kong.” CNBC’s analysts, Dan Wang of Eurasia Group and Han Lin of The Asia Group, talk about proving scale, profitability and regulatory alignment; those quotes are CNBC’s.

Meta’s own consumer-agent work after the unwind is a different product story. The wires mention Muse. We covered a separate Meta agent-protocol note in Sierra and the personal-agent protocol. Workplace agents from another lab are in Google’s Gemini agent at work.

What did Manus 2.0 and Cue add, according to the company?

The English 2.0 post describes a new in-house harness called Cascade, a purchasable Cloud Computer, event-triggered Automations, a desktop workspace named Manus Studio, and a standalone app called Cue. Cascade’s efficiency claims are the company’s: in one tested configuration it used 23.2% fewer tokens, finished tasks in 28.2% less time, and cost 32% less than the previous system. We did not rerun that configuration.

Cue, on that page, gives each personal agent its own email, phone number, wallet and computer, and lets several agents share a group chat. The page says Manus 2.0 is available on web, desktop and mobile; Cue is in early access, free with an invite code, with iOS “coming soon after App Store review.” The sample code printed there is MEETCUE. We did not create an account or use the code.

TechCrunch’s product paragraph matches those buckets: a chatbot and “vibe-coding” tools, Manus 2.0 around a new harness, and Cue as a stand-alone app with email, phone, wallet and computer. Those lines rest on the company pages and on TechCrunch’s summary, not on a test here.

What should readers not treat as confirmed?

Manus did not disclose a valuation. TechCrunch says the company “did not respond to TechCrunch’s questions about its valuation.” CNBC and SiliconANGLE both point to Bloomberg from last month: talks to raise $500 million at a $4 billion valuation. SiliconANGLE adds that $4 billion is “about double what Meta Platforms Inc. had reportedly offered.” That arithmetic is SiliconANGLE’s reading of Bloomberg, not a Manus filing we opened.

We did not open the WeChat post, a Chinese newsroom original, a term sheet, or a Hong Kong listing document. Nikkei Asia’s 8 October headline used a $20 billion Meta figure; that conflicts with the $2 billion used by TechCrunch and CNBC, so we are not using Nikkei as a source. A September TechCrunch story about the then-unclosed round is background, not this week’s announcement.

What did we not see?

This is an evidence review of three 8 October English wires and two undated English Manus blogs opened on 9 October. We did not fetch WeChat, sit in on the round, verify revenue, or run Cascade, Cue or Cloud Computer. Treat the $500 million floor as what the wires say Butterfly Effect said, and treat valuation, ARR and a Hong Kong listing as reported or prior talk.