Meta spent billions building free AI that nobody pays for. Then it bought Manus, a startup that convinced millions to subscribe in eight months. The acquisition reveals more about Meta's monetization struggle than its technology.
AI coding tools promise 55% productivity gains. A new study found developers actually work 19% slower with them—but feel faster. The gap between perception and reality explains why some users can't stop prompting, even at 2 AM.
Nvidia paid $20 billion for Groq's technology and hired its leadership. But it's not an acquisition, according to Jensen Huang. The licensing deal structure has become Silicon Valley's preferred method for absorbing competitors while avoiding regulatory scrutiny.
Meta spent billions building free AI that nobody pays for. Then it bought Manus, a startup that convinced millions to subscribe in eight months. The acquisition reveals more about Meta's monetization struggle than its technology.
Meta spent the year talking up its open-source AI models while burning through billions on infrastructure. Then it acquired Manus, a Singapore startup with Chinese founders that did something Meta hasn't: convinced millions of people to actually pay for AI.
The deal, announced Monday, brings Meta a subscription service that hit $100 million in annual recurring revenue eight months after launch. No other startup globally has reached that milestone faster, according to the company. For context, Meta has been giving away its AI capabilities for free across Instagram, WhatsApp, and Facebook while Zuckerberg promises Wall Street that monetization will eventually materialize. Manus already found it.
Meta says it will keep operating Manus as a paid service while also integrating the technology into its own products. That's the tell. The company isn't just buying Manus's AI agent technology—it's acquiring evidence that the business model works.
Key Takeaways
• Meta acquired Manus for its $100M ARR in 8 months, proving subscription AI agents work while Meta's free AI shows no monetization path
• Second major AI acquisition after Scale AI's $29B deal, following pattern of buying external proof rather than demonstrating internal capabilities
• Manus operates under Chinese parent company, creating awkward optics after lawmakers criticized Benchmark's earlier investment in the startup
• Acquisition contradicts Meta's open-source positioning, as company will continue operating Manus as paid service alongside free Meta AI
Manus offers what it calls a "general-purpose AI agent" that handles research, data analysis, coding tasks, and automation. The company processed 147 trillion tokens and created 80 million virtual computers since launching in March. Behind those numbers: servers humming through endless spreadsheet comparisons, resume scans, stock analyses—the kind of grinding digital work that nobody wants to do manually but also nobody thinks about until it's automated away.
The actual product does things like screen resumes, build trip itineraries, analyze stocks, create websites. Useful tasks, certainly. Revolutionary technology? Less clear. What matters is that people paid for it immediately and kept paying. The company's total revenue run rate now exceeds $125 million when you include usage-based fees beyond the base subscription.
Benchmark led a $75 million funding round for Manus earlier this year at roughly a $500 million valuation. That investment drew sharp criticism from U.S. lawmakers because Manus operates under Beijing Butterfly Effect Technology Ltd Co, a Chinese parent company. Senator John Cornyn called out Benchmark directly: "Who thinks it is a good idea for American investors to subsidize our biggest adversary in AI?" Now Meta owns what Benchmark backed, and the awkwardness hasn't diminished.
This marks Meta's second major AI acquisition in 2025 following a similar playbook. The company took a 49% stake in Scale AI valued at $29 billion, primarily to bring Scale founder Alexandr Wang in as Meta's chief AI officer. That deal wasn't really about Scale's data-labeling business. It was renting a pilot because Meta doesn't trust its own navigation system.
The Manus acquisition follows the same logic. Meta has the researchers, the infrastructure, the distribution. What it lacks is proof that people will pay for AI agents when they can get ChatGPT or Claude for $20 a month. Manus provides that proof along with a built-in user base already comfortable with subscription pricing.
Meta's existing AI offerings remain free across its social platforms. The company has promoted this free access while competitors like OpenAI and Anthropic charge for premium tiers. The strategy makes sense for user acquisition but creates a monetization problem. How do you start charging for something you've been giving away? Easier to buy a company that never made that mistake.
Zuckerberg has pledged $60 billion toward U.S. infrastructure projects over three years, much of it AI-related. The company assembled an expensive research team to develop new models planned for spring 2026. All that spending needs to generate revenue eventually, and Meta hasn't demonstrated a clear path beyond better advertising targeting.
Manus represents a different approach: charge users directly from day one, prove the value proposition, scale revenue before worrying about market dominance. Meta buying this proves the subscription model it hasn't proven internally.
Meta has cast itself as the champion of open-source AI development. The company released its Llama model series with permissive licensing, arguing that widespread access accelerates innovation and prevents monopolistic control. Zuckerberg has explicitly contrasted Meta's approach with OpenAI's closed, profit-driven model.
Now Meta is acquiring a paid subscription service and explicitly plans to keep operating it that way. The company says it will "continue to operate and sell the Manus service" while also integrating it into Meta's products. That's a departure from Meta's previous pattern of absorbing acquisitions into free offerings.
Strategic AI news from San Francisco. Clear reporting on power, money, and policy. Delivered daily at 6am PST.
No spam. Unsubscribe anytime.
The contradiction betrays a specific insecurity: Meta trusts its code, but it doesn't trust its ability to sell it. Open-source positioning always had a strategic element—giving away AI models costs Meta less than it costs smaller competitors without comparable infrastructure. But when Meta needs to prove monetization, it buys a closed, paid service rather than building one internally.
Manus proved in eight months what Meta couldn't in a year. The difference shows up in the friction. Free products disappear into your social feed—you use them without thinking, without pulling out a credit card, without deciding if they're worth $20 versus alternatives. Paid products force that decision every month. People chose to pay Manus, repeatedly, which means the product cleared a bar Meta's free AI hasn't: it convinced users it was worth money they could spend elsewhere.
Meta's AI strategy has relied on a fundamental assumption: build the best free AI, integrate it everywhere, monetize through better ad targeting and eventual premium features. That works if you can demonstrate that premium features will actually generate meaningful revenue. Meta hasn't shown that yet with its own products.
The China connection creates uncomfortable optics but probably won't derail the deal. Manus relocated to Singapore, and Meta will continue operating it from there. The technology itself isn't particularly novel—the business model execution is what matters. If lawmakers objected to Benchmark's investment, they may object to Meta's acquisition, but stopping the deal would require proving specific security concerns beyond general China-origin suspicion.
The broader pattern is more significant than this single deal. Meta keeps acquiring external proof rather than demonstrating capabilities internally. Scale AI for labeling credibility and talent. Manus for subscription revenue confirmation. These deals suggest impatience about Meta's ability to execute certain strategies from scratch despite enormous resource advantages.
For a company spending $60 billion on AI infrastructure, buying a $125 million revenue run-rate startup shouldn't be strategically necessary. It should be a rounding error compared to internal execution. That Meta made this acquisition anyway indicates real questions about where the money is coming from.
The deal might work out well. Manus gets Meta's distribution and resources. Meta gets proven revenue streams and a user base already paying for AI agents. But the acquisition also exposes what Meta couldn't build internally: a profitable AI product that people chose to pay for before being offered a free alternative.
Wall Street wants to see AI spending translate into revenue growth. Meta just bought someone else's answer to that question rather than developing its own.
Q: What's the difference between an AI agent and a chatbot like ChatGPT?
A: Chatbots respond to prompts and stop. AI agents execute multi-step tasks without constant supervision—they can research a topic across multiple sources, compile findings, create deliverables, and iterate on work autonomously. Manus's agent can build entire websites or analyze stocks without users guiding each step. The difference: chatbots answer, agents act.
Q: How much did Meta pay to acquire Manus?
A: Meta didn't disclose the purchase price. Manus raised $75 million at a $500 million valuation in early 2025, so the acquisition likely commanded a premium above that. Given the $100M ARR milestone and strategic value to Meta, analysts estimate the deal could range from $800 million to $1.5 billion, though this remains speculation.
Q: How did Manus grow to $100M ARR in just 8 months?
A: Manus launched in March 2025 following DeepSeek's release, which created massive interest in Chinese AI capabilities. The company offered free task completion on X to demonstrate capabilities, then converted users to paid subscriptions. Processing 147 trillion tokens across 80 million virtual computers suggests aggressive user acquisition combined with high retention on subscription pricing.
Q: Will Manus remain a separate paid product or merge into Meta AI?
A: Meta explicitly said it will "continue to operate and sell the Manus service" while also integrating the technology into Meta's products. This suggests a dual strategy: keep Manus as a premium subscription offering for business users while incorporating agent capabilities into free Meta AI across Instagram, WhatsApp, and Facebook.
Q: Why does Manus's Chinese connection create problems for Meta?
A: Manus operates under Beijing Butterfly Effect Technology Ltd Co, raising concerns about potential data access by Chinese authorities. Senator John Cornyn criticized Benchmark's investment for "subsidizing our biggest adversary in AI." While Manus relocated to Singapore, the Chinese parent company structure could trigger regulatory scrutiny or congressional pressure on Meta's acquisition, especially for a product handling business research and sensitive data.
Marcus Schuler
Maria Garcia
Marcus Schuler
AI coding tools promise 55% productivity gains. A new study found developers actually work 19% slower with them—but feel faster. The gap between perception and reality explains why some users can't stop prompting, even at 2 AM.
Nvidia paid $20 billion for Groq's technology and hired its leadership. But it's not an acquisition, according to Jensen Huang. The licensing deal structure has become Silicon Valley's preferred method for absorbing competitors while avoiding regulatory scrutiny.
Astra AI just won Slovenia's Startup of the Year. The numbers look great: 170,000 users, expansion into Germany, rave reviews. But their claimed 50 billion monthly tokens raises a question no one's asking: can a €24/month subscription cover those API bills?
Zhipu AI's GLM-4.7 benchmark chart excludes its strongest competitor. The data table tells a different story. But the real signal isn't the missing bar—it's a €30 annual subscription designed to snap into tools Western labs built.
Get the 5-minute Silicon Valley AI briefing, every weekday morning — free.
