Meta Buys 9-Month-Old AI Agent Startup Manus for Over $2 Billion
Meta has agreed to acquire Manus, a Singapore-based developer of general-purpose AI agents, in a deal valued at more than $2 billion. The transaction marks one of the fastest zero-to-exit sprints in enterprise software history, occurring just nine months after the startup publicly launched its flagship product.
- Singapore-based Manus hit $100M ARR in eight months and is currently on a run rate north of $125 million.
- The acquisition removes all Chinese ownership from the cap table while vaulting Meta into the lead for autonomous AI agents.
The deal serves as a stark rebuttal to the traditional venture timeline that suggests building a unicorn requires a decade of grind and a massive headcount. Instead, Manus capitalized on a lean team and immediate product-market fit to secure a multi-billion dollar outcome in under a year.
“Manus is already serving the daily needs of millions of users and businesses worldwide,” Meta stated regarding the deal. The company confirmed that Manus’s general-purpose agents will be integrated into Meta AI across Facebook, Instagram, and WhatsApp, giving the social giant a functional layer to sit on top of its Llama models.
Manus CEO Xiao Hong will join Meta as a vice president, reporting directly to the company’s AI leadership. He framed the acquisition not as an off-ramp, but as an acceleration.
Joining Meta allows us to build on a stronger, more sustainable foundation without changing how Manus works or how decisions are made.
From Beijing to Singapore to Menlo Park
Manus is not a typical Silicon Valley story. The company traces its lineage to Butterfly Effect (also known as Monica.im), a startup originally founded in Beijing. The team quietly shifted its headquarters to Singapore earlier this year, a strategic maneuver to navigate rising geopolitical tensions and investment restrictions from Washington.
From that new base, Hong and a team of roughly 100 engineers focused on a specific thesis: chatbots are table stakes, but “virtual colleagues” that can execute complex workflows are the real prize.
Manus’s agents go beyond conversation. They screen résumés, analyze financial portfolios, and handle back-office logistics by chaining together multi-step tasks. In 2025, Microsoft tested Manus on Windows 11 to help users build content directly from local files, proving the utility extended far beyond a browser tab.
Insiders note that the pivotal moment came when early adopters stopped treating the software like a toy and started using it as a junior analyst. Once power users began running core business functions on the agents, revenue compounded immediately.
The $100M Sprint
Manus publicly launched its general-purpose agent earlier this year. Within eight months, it crossed $100 million in annual recurring revenue (ARR), driven largely by subscriptions and heavy commercial usage.
Sources close to the company confirm the startup is now sitting on a run rate exceeding $125 million. This was achieved with a headcount of around 100 employees—all of whom will move to Meta—resulting in a revenue-per-employee metric that is practically unheard of in modern SaaS.
The team bootstrapped its internal culture while taking capital from heavy hitters like Tencent, ZhenFund, and HSG (formerly Sequoia China). They prioritized GPU capacity and product velocity over sales teams or brand marketing. The product effectively sold itself.
In a world of AI decks and zero revenue, Manus had something radical: millions of paying users and agents people actually trusted with work.
Betting on Agents, Not Just Chatbots
This is not a standard “acqui-hire” to grab talent. For Meta, this is its third-largest acquisition ever, trailing only WhatsApp and its strategic stake in Scale AI. It follows a massive $14.3 billion investment in Scale and the formation of Meta Superintelligence Labs.
The strategic logic is straightforward: Large Language Models (LLMs) like Llama provide the intelligence, but they need an execution layer to actually perform tasks. Manus provides that plumbing, allowing Meta to potentially replace a swath of third-party SaaS tools within its own ecosystem.
Under the terms of the deal, Manus will continue to operate as a standalone product while its underlying tech is wired into Meta’s business offerings. This gives Meta an immediate subscription revenue stream and a toolset to drive deeper engagement.
A Clean Break from China
The deal structure is heavily influenced by the current geopolitical climate. Meta has made it explicitly clear that there will be “no continuing Chinese ownership interests in Manus AI” once the deal closes. The startup will discontinue services and operations inside China, severing ties to its original market.
This clean break is designed to satisfy regulators in Washington who are increasingly scrutinizing cross-border AI transactions. Existing backers, including the Chinese VCs who funded the original Butterfly Effect entity, will be cashed out. Early employees are looking at life-changing payouts less than three years after the project began.
For Meta, buying a proven, revenue-generating agent is likely cheaper and faster than trying to build one in-house while OpenAI and Google race ahead. For Hong, it trades a risky, capital-intensive path to a potential IPO for the safety and distribution of Meta’s balance sheet.
A New Startup Playbook
The Manus exit effectively rewrites the pitch deck for the AI era. The company went from obscurity to a ten-figure exit without the traditional bloat of a decade-long roadmap or a 500-person office.
It signals a shift in venture capital expectations. In the age of agents, the valuation premium has shifted from having a flashy model demo to having a product that can turn intelligence into immediate income. Manus proved that if the agents work, you don’t need ten years to build a billion-dollar business.
