NextFin News - Manus, the AI agent startup at the center of a collapsed $2 billion takeover by Meta Platforms, is in talks for its first funding round since the deal unraveled — and the price tag has doubled. The company is targeting a valuation of roughly $4 billion, according to people familiar with the matter, a figure that would value the Singapore-headquartered business at twice what Meta agreed to pay just nine months ago and eight times the roughly $500 million benchmark set in its April 2025 Series B.
The round is the clearest signal yet that the breakup of the Meta-Manus tie-up has not dimmed investor appetite for autonomous AI agents. It also raises the stakes for a company that must now prove its estimated $450 million to $500 million in annualized revenue can compound outside the distribution machine of the world's largest social media company.
The Deal That Was, and the Round That Is
Meta agreed to acquire Manus in December 2025 at a reported valuation of more than $2 billion. The transaction was structured to keep Manus operating from Singapore while winding down its China operations, with founder and CEO Xiao Hong moving into Meta's leadership structure. By December 17, 2025, the company said it had crossed $100 million in annual recurring revenue eight months after launch — a pace it called the fastest in the world from zero to that milestone — with a total revenue run rate above $125 million.
Beijing saw the transaction differently. On April 27, 2026, China's National Development and Reform Commission ordered Meta to unwind the acquisition, citing national security concerns and the imperative of keeping leading AI companies under domestic control. The regulator said the decision to prohibit foreign investment in Manus was made in accordance with applicable laws and asked both parties to withdraw the transaction. Meta completed the operational split by June 11, 2026: data sharing stopped, Manus staff were cut off from Meta's internal systems, and Meta employees were barred from using Manus tools internally.
What followed was a scramble to reconstruct ownership. Early Chinese backers — Tencent Holdings, HongShan (formerly Sequoia China), and ZhenFund — moved to buy back the company from Meta at roughly the same $2 billion valuation Meta had paid, according to a June 18, 2026 report. The structure under discussion would leave Manus organized as a Chinese joint venture with those investors, positioning it for a potential Hong Kong initial public offering. In parallel, the company's founders explored raising about $1 billion of their own to help finance the buyback, with personal capital on the table.
Now, with a buyback consortium led by Tencent in place, the company is testing whether the market will pay a premium for a second chance. The roughly $4 billion target would represent an eightfold step-up from the $75 million Series B led by U.S. venture firm Benchmark. Benchmark, which was paid out in the Meta transaction, is not part of the new consortium.
Why $4 Billion Is Not Just a Meta Hangover
The valuation math is unforgiving, and it cuts both ways. Against estimated annualized revenue of $450 million to $500 million as of June 2026, a $4 billion price tag implies roughly eight to nine times revenue. The ~$2 billion buyback price, by contrast, sits at about four times that same revenue base. The gap between the two is a bet that growth, not just the asset, is being priced.
There is a case for the premium. Manus was never built as a feature inside someone else's product. It is an autonomous agent platform: a user describes a task in plain language, and the system spins up a dedicated cloud virtual machine equipped with browsers, code interpreters, office applications, and design engines to execute it. It can research markets, write code, build spreadsheets, and generate mobile applications without traditional development tools. That independence is precisely what made it valuable to Meta — and what makes it viable on its own.
The company's pricing power also argues for the higher valuation. Manus introduced paid subscription plans in 2025, with tiers priced up to $199 per month for premium usage, layered over a freemium funnel that reportedly generated a waitlist of more than 2 million users before monetization began. That is a consumer-software profile, not a research-lab curiosity. By June 2026, annualized revenue had reached an estimated $450 million to $500 million, up from $127 million earlier in the year — a more than threefold increase in roughly six months, according to revenue trackers.
But the revenue figure deserves scrutiny. Annual recurring revenue in the AI agent category is a notoriously elastic metric: there is no standardized calculation, and usage-based credits can be counted differently depending on whether they are sold, consumed, or merely allocated. The company's own December 2025 disclosure drew a distinction between $100 million in ARR and a $125 million total revenue run rate that included usage-based and other revenue — a gap that suggests a meaningful share of the top line comes from non-recurring consumption. Investors paying $4 billion will want to know how much of that revenue renews automatically and how much evaporates if users stop running expensive multi-step tasks.
The Second-Order Question: Who Actually Owns the Moat?
Beneath the valuation debate lies a harder question about where Manus's competitive advantage actually lives. The platform is built on top of Anthropic's Claude language models, augmented with the company's proprietary context-engineering techniques — the scaffolding that breaks complex tasks into steps, manages memory across long sessions, and decides when to call external tools. Manus 1.5, released in October 2025, cut average task completion time from 15 minutes to under four minutes; a later update added a Design View and lifted user satisfaction by more than 19 percent in double-blind testing, according to company data.
That is real engineering. But it is engineering layered on rented intelligence. Every improvement in Claude's native tool-use and computer-use capabilities narrows the gap between what Manus orchestrates and what the base model can do alone. The moat, in other words, is a process advantage, not a model advantage — and process advantages compress faster than model advantages in a market where Anthropic, OpenAI, and Google are all racing to make their models better agents.
This is the structural tension the $4 billion price tag must resolve. If Manus is a cyclical winner — a hot product riding the 2025-2026 agent hype cycle — then the valuation peaks before mean reversion sets in, as model providers internalize agent orchestration and competition from well-funded rivals pushes pricing down. If Manus is a structural winner — a durable workflow layer with switching costs, enterprise integrations, and proprietary task-execution data that compounds over time — then $4 billion could look cheap in hindsight.
The evidence points to a hybrid: a structural opportunity wrapped in a cyclical valuation. The structural leg is distribution — millions of users, a paid base, and a partnership that makes Manus a launch partner for enterprise agent workflows inside Microsoft 365. The cyclical leg is the multiple itself, being set in a market where Chinese AI valuations have been bid up by a scarcity of investable crown jewels. More than 85 percent of Chinese AI-related companies that went public through mid-2026 chose Hong Kong over Shanghai or Shenzhen, according to a Kharon analysis, and recent listings from Zhipu AI and MiniMax have shown that the market will absorb large raises even after steep run-ups. Manus is pricing itself into that same window.
The Competitive Landscape: A Category Still Being Invented
Manus is not raising in a vacuum. The autonomous-agent category it helped popularize has become one of the most crowded and best-funded corners of artificial intelligence. Genspark, which builds the same category of general-purpose agent product, expanded its Series B to $385 million at a $1.6 billion valuation by April 2026, backed by Emergence Capital, LG Technology Ventures, SBI Investment, and Pavilion Capital. OpenAI and Google have both moved to bake agent capabilities — tool use, browser control, long-horizon task execution — directly into their base models. Microsoft has made agents the centerpiece of its enterprise software roadmap.
That convergence cuts two ways. On one side, it validates the category: the largest technology companies in the world are betting that the next interface layer of computing is not a chatbot but an agent that acts on the user's behalf. On the other, it means Manus's window to establish a defensible position is narrowing. A process advantage built on top of third-party models is exactly the kind of advantage that gets arbitraged away when the underlying model providers decide to compete directly. The company's partnership with Microsoft, which makes its workflows run natively inside Microsoft 365, is therefore more than a distribution deal — it is a hedge against being disintermediated by the platform owners.
The capital markets backdrop helps explain why a $4 billion round is plausible even with those risks. Chinese technology companies have turned to Hong Kong's markets to fund research, hiring, and expansion as China seeks to narrow the gap with the United States in artificial intelligence. More than 85 percent of Chinese AI-related companies that went public through mid-2026 listed in Hong Kong rather than Shanghai or Shenzhen, according to a Kharon analysis, and the combined Hong Kong-Shanghai IPO market raised more than $54 billion in 2026, a 17 percent increase from the prior year, according to LSEG data. In that environment, a proven agent company with hundreds of millions in revenue and a repatriated ownership structure is a scarce asset — and scarcity, not just fundamentals, sets private-market prices.
But the same backdrop contains the seeds of the downside case. Hong Kong listings in 2026 have been marked by steep first-day pops followed by heavy discounts to private-market valuations as the year progressed. Zhipu AI priced a follow-on share sale at a discount even after its stock had surged more than 2,000 percent since its January 2026 debut. If Manus follows the same path to public markets, the investors paying $4 billion today may find themselves marking the company down before it ever trades. The round is, in that sense, a race against the public-market window — and windows close faster than private rounds price.
The Counter-Case: A $2 Billion Company Wearing a $4 Billion Suit
The strongest argument against the new round is simple: Manus already had a bid above $2 billion, and the buyer walked away — not because it changed its mind about the technology, but because Beijing would not let the deal close. That bid was set against roughly $100 million in ARR. Paying $4 billion against up to $500 million in annualized revenue is not a vote for the technology; it is a vote that revenue grew roughly fivefold in six months and will keep compounding. If growth slows — if post-breakup churn, the loss of Meta's internal usage, or a broader AI spending slowdown bites — the multiple has nowhere to go but down.
There is also the question of what a new owner actually receives. Manus's technology was being woven into Meta's infrastructure during the months between announcement and unwind. Separating the two is not clean: user data created during the acquisition window was slated for deletion, and the company warned that some user accounts would be affected by the return to independence. A buyer at $4 billion is not just buying a revenue stream; it is buying a reconstruction project.
The falsifying signal is quantifiable: if Manus's annualized revenue does not hold above $400 million through the first half of 2027 — or if net revenue retention falls below 100 percent as measured in a future disclosure — the $4 billion valuation loses its growth justification and reverts toward the $2 billion level Meta actually agreed to pay. Conversely, if the company can demonstrate that enterprise usage through Microsoft 365 is accelerating and that subscription revenue, as distinct from usage credits, is growing, the premium holds.
What Comes Next
The round, if completed, would set up three distinct timelines. In the short term, the news itself is a sentiment catalyst: it tells the market that Manus survived the geopolitical rupture and that Chinese capital is willing to pay up for AI assets that can be repatriated onshore. In the medium term, the money buys runway to prove the revenue model outside Meta — specifically, whether the Microsoft partnership and the consumer subscription tiers can replace the distribution the company gave up. In the long term, the joint-venture structure and the Hong Kong IPO path position Manus as a rare cross-border AI asset: Chinese-controlled, Singapore-headquartered, and listable in the market that has become the default venue for Chinese AI capital.
Scenarios split cleanly. The base case: the round closes near $4 billion, the Tencent-led consortium takes control, and Manus files for a Hong Kong listing within the next 12 to 18 months at a valuation that reflects public-market discounts to private AI multiples. The upside case: revenue continues to compound at a rapid pace, the Microsoft integration deepens, and Manus becomes the reference listing for Chinese AI agents, commanding a premium to recent peers. The downside case: growth decelerates, the separation from Meta proves costlier than expected, and the company reprices toward the $2 billion buyback level, leaving new investors underwater.
The watchlist is short and specific: the final size of the round and the identity of any new institutional investors beyond the existing consortium; the next revenue disclosure, with particular attention to the split between subscription ARR and usage-based revenue; and any filing that confirms the Hong Kong IPO track.
"Manus will soon return to operating as an independent company, which will see us continue to serve our millions of users around the world," the company said in an August 11, 2026 statement, adding that "during this process, it is inevitable that some user accounts will be affected."
The central judgment: $4 billion is not a valuation of what Manus is today — a revenue business freshly severed from its would-be parent. It is a bet that the AI agent category will still be bidding for scarce, proven winners when the company reaches public markets, and that Beijing's willingness to block a U.S. takeover has turned a Singapore startup into a domestic crown jewel worth paying a premium to own. If the category cools before the IPO, the premium evaporates first.
Explore more exclusive insights at nextfin.ai.