Meta 砸 20 億美元買下 Manus:這筆 10 天敲定的收購,把「AI 員工」推上檯面

At the end of last December, on a certain workday, a friend who works on SaaS in Taipei sent me a news link with a title that roughly meant "Meta to Buy Manus". He added a line: "You often use agent tools, and this company is about to be acquired." I opened the link and was taken aback - not because of the acquisition, but because of the amount and speed: over $2 billion, with negotiations reportedly taking only about ten days.

Having observed the enterprise world for several years, I've seen many AI acquisition cases, but the combination of "ten days" and "two billion" still made me sit up straight. This isn't a strategic layout made at leisure; it's more like someone is rushing to grab something, and they're in a hurry.

Event Background

Manus is a company registered in Singapore, founded by a team with a Chinese background, and it develops "general-purpose AI agents" - you give it a goal, and it will break it down into a series of sub-tasks, then use its built-in tools to complete the task step by step, rather than just providing a text response. Its target customer group is clear: small and medium-sized enterprises that can't afford to hire a whole team but want to use AI to supplement their workforce.

According to reports from Bloomberg and CBC, after Manus went public, its annual recurring revenue (ARR) reached the $100 million level in just about eight months. For a company so young, this growth rate is a signal in itself: the market is willing to pay for "AI that can do things", not just "AI that can chat".

Key Points

According to reports from Bloomberg and CBC, the key information about this transaction is as follows:

• Amount: valued at over $2 billion, another multi-billion dollar bet by Meta on AI.
• Time: announced on December 29, 2025, and finalized in early 2026, with negotiations completed in about ten days.
• Target: Manus, a Singapore-registered company with a Chinese background team, focusing on autonomous AI agents for small and medium-sized enterprises.
• Achievement: reached an ARR of $100 million in about eight months after going public.
• Significance: Meta has previously invested in "models" and "computing power", but this time it directly acquired an "agent that can take action", filling in the missing piece of "hands" in its AI landscape.

Market Impact Analysis

For Taiwanese users: in the short term, you won't feel much impact from this acquisition. However, the direction is worth noting - when a company like Meta recognizes agents as the next step, it means you'll encounter more "AI that can complete tasks" in various apps, social media, and work software, rather than just "AI that answers questions". It's a good idea to get familiar with the operating logic of tools like Manus ahead of time.

For enterprise applications: this is a reminder for Taiwanese small and medium-sized enterprises. Manus succeeded because it met the demand for "one person to handle multiple tasks", which is similar to the situation faced by many of our small and medium-sized enterprises. There are already tools like Tycoon AI on the market that claim to provide "a team of AI employees", although they're not yet mature enough to replace human teams. It's worth starting to test the waters on a small scale, rather than waiting until your competitors are already using them.

For developers: the recognition of agents by giants means that there will be more resources and opportunities for this technical route. Whether it's developing applications based on agent frameworks or researching platforms like Google Antigravity that Google has just launched, now is the time to invest time and effort.

Future Development Trends

From 2025 to 2026, the main battlefield of AI has clearly shifted from "who has the smarter model" to "whose agent can complete tasks better". Meta's acquisition of Manus, Google's launch of Antigravity, and various companies competing to develop multiple agent collaborations - all of these are based on the same consensus: chatbots are just the starting point, and agents that can complete tasks autonomously are where the money will really flow. However, it's also important to remain calm: the reliability, safety, and responsibility of agents' autonomous execution are still not fully resolved, and there is still a distance between the hype and maturity.

TheAI Academy Summary and Comments

What Taiwanese readers should remember most about this acquisition is not the $2 billion figure, but the fact that it confirms one thing: the market is willing to pay for "AI that can do things".

Comments: don't just focus on how much money the giants are spending; ask yourself if your company has started using similar technologies. Agents are not yet mature enough to be left unattended, but if you don't start learning now, you'll be left behind when they do mature.

Practical advice for Taiwanese readers: start by selecting a repetitive and non-critical workflow (such as data organization, preliminary research, or drafting responses), and use agent tools to test it out. Treat the agent as an intern to be guided, rather than an employee to be trusted. This article involves enterprise mergers and acquisitions and market judgment, and is for reference only, not constituting investment advice.

Data Sources

• Bloomberg: Meta to Buy Manus, an AI Startup With Chinese Roots
• CBC News: Meta acquires Chinese-founded AI startup Manus for $2B

(This article is compiled based on publicly available information, with numbers and times verified against official and original reports.)