Meta’s Boldest AI Bet Yet: Why Zuckerberg Just Spent $2B on a China-Born AI Agent That Could Reshape the Tech Arms Race

Meta Platforms’ agreement to acquire AI start-up Manus marks one of the most consequential—and symbolically loaded—AI deals of the year, underscoring just how aggressively Mark Zuckerberg is pushing the company into the next phase of artificial intelligence. Valued at more than $2 billion, according to people familiar with the transaction, the deal gives Meta an immediate foothold in the fast-emerging market for autonomous AI agents while also raising unavoidable questions about geopolitics, regulation, and the future shape of the AI ecosystem.

At its core, the transaction is straightforward. Meta is buying Manus outright, including buying out all existing investors, and plans to continue operating the product as a subscription service while integrating its capabilities across Meta’s platforms. The Wall Street Journal first reported that the deal exceeded $2 billion, with other sources placing the value closer to $2–3 billion. Manus was in the process of seeking new funding at a roughly $2 billion valuation when Meta approached, accelerating negotiations that reportedly concluded in about ten days. Manus CEO and co-founder Xiao Hong is expected to report to Meta COO Javier Olivan, while the roughly 100-person team will be folded into Meta’s broader AI organization.

What Meta is buying is not a foundation model, but a highly capable AI agent layer built on top of existing models from companies such as Anthropic and Alibaba. Manus burst onto the scene in March after previewing an autonomous agent capable of taking simple instructions and executing multi-step tasks with minimal human supervision. The system can generate in-depth research reports, build custom websites, analyze data, write and debug code, and coordinate complex workflows—functions that go well beyond traditional chatbots. That positioning helped Manus gain millions of users in a matter of months, including a growing base of paying subscribers.

The commercial traction is what truly set Manus apart. By December, just eight months after launch, the company reported crossing $100 million in annual recurring revenue, with some estimates placing its revenue run rate closer to $125 million. That kind of growth is rare even by AI standards and gives Meta something it has lacked in its AI push so far: a proven, monetized AI product that users are already paying for directly. Manus also ranked as the top-performing AI on Scale AI’s Remote Labor Index, a benchmark designed to assess how well AI agents perform real-world digital work—an endorsement that likely resonated strongly with Meta’s leadership.

Strategically, the acquisition fills a clear gap in Meta’s AI portfolio. Meta has invested heavily in foundational capabilities—custom chips, massive data-center buildouts, and open-source Llama models—but it has lagged rivals like OpenAI, Microsoft, and Google in delivering high-end, task-oriented AI agents. Microsoft has Copilot embedded across Windows and Office, OpenAI is expanding ChatGPT into enterprise workflows, and Google is weaving Gemini into Workspace. Until now, Meta’s AI presence has skewed toward consumer chatbots embedded in Instagram, WhatsApp, and Facebook, along with ad-optimization tools that quietly boost margins. Manus gives Meta a credible entry point into agentic AI for small businesses and professionals, particularly within WhatsApp’s rapidly growing SMB ecosystem.

The timing is also notable. Zuckerberg has made AI the company’s top strategic priority, committing to spend as much as $600 billion in the U.S. through 2028 if AI progress continues to accelerate. Meta recently launched a new Superintelligence Labs division and recruited Alexandr Wang, founder of Scale AI, after acquiring a 49% stake in his company for $14 billion. Against that backdrop, a $2-plus-billion acquisition for a fast-growing AI agent platform looks almost modest—yet potentially transformative. Rosenblatt analysts went so far as to suggest the Manus deal could eventually stand alongside Meta’s acquisitions of Instagram and WhatsApp in terms of long-term impact.

Still, the deal is not without risk, particularly on the regulatory front. Manus has Chinese roots, having emerged from a start-up called Butterfly Effect that was founded in China in 2022 with offices in Beijing and Wuhan. Early development relied heavily on Chinese engineering talent, and the company initially garnered attention as a symbol of China’s AI capabilities, especially following the debut of DeepSeek. That background alone is enough to trigger scrutiny in Washington, especially at a time when U.S.–China technology tensions remain high.

Meta has clearly anticipated those concerns. Manus moved its headquarters to Singapore earlier this year after raising $75 million in a funding round led by Benchmark, and it has since shut down its China operations and shelved plans to offer its product in the Chinese market. Meta has explicitly stated that there will be no continuing Chinese ownership interests after the transaction and that Manus will discontinue all services in China. All existing investors, including Tencent, ZhenFund, and HSG, are being bought out. Those steps are designed to reduce national-security concerns, but they may not eliminate them entirely. U.S. regulators could still examine data flows, model training practices, and historical ties as part of any review.

That said, the regulatory risk appears manageable relative to other high-profile tech deals. Manus does not control critical infrastructure, does not produce foundational AI models tied to sensitive hardware, and is now legally domiciled in Singapore with operations focused outside China. The Committee on Foreign Investment in the United States could still take interest, but Meta’s proactive restructuring likely lowers the probability of a prolonged or blocking review.

For Meta shareholders, the market reaction has been constructive but restrained. Shares rose modestly following the announcement, extending a year-to-date gain of roughly 13%, though still trailing the broader S&P 500. That response reflects both enthusiasm for the strategic fit and recognition that Meta is already deeply committed to AI spending. The bigger question is execution: whether Meta can successfully integrate Manus’ subscription-driven, agentic capabilities into its ecosystem without diluting what made the product successful in the first place.

In many ways, the Manus acquisition signals where the AI race is heading next. The era of ever-larger models is giving way to competition over who can turn intelligence into action—AI that doesn’t just talk, but acts. For Meta, buying Manus is a bet that the future of AI engagement lies not only in social feeds and ads, but in autonomous digital workers embedded across everyday business and consumer workflows. Whether that bet pays off will depend on how effectively Meta can scale Manus globally—and how smoothly it can navigate the geopolitical crosscurrents that come with acquiring one of the most prominent AI products to emerge from Asia’s startup ecosystem.