On December 30, 2025, social media giant Meta announced it had acquired Manus, a Chinese AI company, and its parent firm Butterfly Effect, for several billion dollars.
This marks Meta’s third-largest acquisition in its history.
Under the agreement, the entire Manus team has joined Meta, with founder Xiao Hong appointed as Vice President overseeing AI agent technology. Manus will continue to operate independently from Singapore, with its products and services remaining unchanged.
01
Following the announcement, observers naturally interpreted the deal through the lens of U.S.-China tech competition.
For the U.S. and Meta, the intent is clear.
Mark Zuckerberg has declared AI Meta’s top strategic priority, investing $54 billion in AI infrastructure in 2025 alone.
Yet, despite massive investments, Meta has lacked a standout, consumer-facing general-purpose AI application—until now.
Manus fills that critical gap.
Its product is the world’s first general-purpose AI agent, featuring a unique technical architecture and achieving an 86.5% accuracy rate on the GAIA benchmark—surpassing OpenAI’s comparable offering.
Alongside DeepSeek, Manus was widely regarded as one of the two most remarkable breakthroughs in China’s AI sector in 2025.
By acquiring Manus, Meta instantly gains a crucial capability: enabling AI not just to answer questions, but to autonomously execute complex tasks.
With access to Meta’s ecosystem of nearly 4 billion monthly active users, this technology could be rapidly scaled to automate content, advertising, and even broader aspects of daily life.
Of course, such a high-profile acquisition—especially of a company with core Chinese origins—is likely to trigger scrutiny. The U.S. Federal Trade Commission may launch an antitrust review, and members of Congress could raise concerns about data security and technological competitiveness.
02
From China’s perspective, reactions are more complex.
On the positive side, it’s undeniable: a young Chinese team built a world-class AI product in mere months and earned recognition—and billions of dollars—from one of the world’s leading tech giants.
This validates the talent and ingenuity of China’s AI innovators.
For early investors like ZhenFund, it represents a highly successful exit, which will likely encourage more capital to flow into AI startups.
Manus’s journey—from founding in China, relocating to Singapore, and ultimately being acquired by a U.S. tech giant—offers a real-world case study for future entrepreneurs navigating today’s intricate global environment.
But there’s another, more sobering dimension.
After the acquisition, the entire core technical team, led by Xiao Hong, moved to Meta—reigniting debates about the trajectory of China’s top AI talent.
Even more telling: the acquired entity wasn’t technically a Chinese company at all. It had long since relocated its headquarters and primary operations to Singapore—a detail that recalls the controversial move earlier in 2025 that sparked widespread discussion.
03
For Manus itself, the Meta acquisition may not have been a surprise, but rather the likely outcome of a path set in motion the moment it chose to move to Singapore.
In hindsight, that relocation was driven by multiple converging pressures.
The most immediate push came from international politics.
In April 2025, after raising $75 million in its Series B round, Manus triggered a national security review by the U.S. Treasury Department under the newly enacted “Outbound Investment Security Program.”
Benchmark Capital, the Silicon Valley firm leading the round, explicitly required the company to reincorporate in Singapore to proceed with funding.
This wasn’t unprecedented. AI video startup HeyGen had previously followed the same playbook: dissolving its Chinese entity, relocating to Los Angeles, and successfully securing Benchmark investment.
Manus simply replicated the model—but chose Singapore instead.
Technical constraints posed another major hurdle.
U.S. export controls on high-end AI chips made it extremely difficult for Manus to reliably access essential computing power like NVIDIA’s H100 within China.
Insiders revealed that insufficient compute capacity had already begun slowing product development.
Singapore, as Asia’s compute hub, offered “compute freedom”—unaffected by U.S. sanctions and capable of supporting intensive R&D.
Deeper still were capital logic and global strategy.
Butterfly Effect—the Singapore-based entity—was actually incorporated as early as August 2023, wholly owned by a Cayman Islands parent company.
This pre-emptive structure, established well before the product gained fame, shows that globalization was a deliberate strategy, not a reactive pivot.
A narrative of “headquartered in Singapore, serving the world” was far more compelling to global investors than “a Chinese AI startup”—helping justify its $2 billion valuation and aligning with the standard Cayman-based exit model favored by U.S. dollar funds.
Domestic market realities also played a role.
Manus’s core technology heavily relied on foreign large models like GPT-4 and Claude, which integrated poorly with domestic Chinese alternatives.
Moreover, compared to the cutthroat price wars in China’s enterprise software market, Western markets showed stronger willingness to pay for premium AI tools—critical for offsetting Manus’s high operational costs.
Thus, from U.S. regulatory scrutiny and compute shortages to investor demands and market dynamics, multiple gears interlocked to steer Manus toward Singapore—and ultimately into the arms of a global tech giant.
When founder Xiao Hong described the move as “building a truly global product under new conditions,” he was articulating the difficult choices a startup must make to survive and thrive amid geopolitical fault lines.
04
The full arc of Manus—from inception, relocation, to acquisition—acts like a spotlight, illuminating deep challenges facing China’s AI industry and broader innovation ecosystem.
First, the issue of talent and ecosystem.
Manus’s team earned global acclaim, yet their成果 (achievements/results) now enrich a foreign tech giant.
This raises a fundamental question: How can China build an environment that not only cultivates top talent but also retains them long-term—so their innovations take root and grow into forests on home soil?
Second, the tension between technological autonomy and globalization.
Much of Manus’s success stemmed from its ability to freely leverage the world’s most advanced large models.
Part of its motivation for relocating was to avoid compliance risks associated with using unapproved foreign models in China.
This highlights a core dilemma: as China pursues AI self-reliance, how can it simultaneously maintain openness and global interoperability?
Balancing security, autonomy, openness, and integration in today’s fast-moving, co-opetitive AI landscape demands exceptional strategic wisdom.
Third, the relationship between capital and innovation.
U.S. dollar capital played a pivotal role in Manus’s early growth—but its exit expectations and compliance requirements directly shaped where the company chose to incorporate.
This forces us to ask: How can China foster more “patient capital”—investors with long-term vision who can accompany hard-tech companies through volatility without fleeing at the first sign of geopolitical turbulence? And how can domestic capital markets develop more attractive exit channels?
Fourth—and perhaps most fundamentally—the resilience of the entire innovation system.
The Manus case reveals a pattern: brilliant entrepreneurs nurtured in China achieve rapid success, yet under the pull of global capital, technology, and markets, they shift their operational centers—and ultimately their value creation—elsewhere.
Is this an isolated incident, or a potential trend?
It compels a thorough examination: Is China’s innovation-supporting ecosystem—from autonomous access to compute infrastructure, to sensible rules on cross-border data flows, to regulatory tolerance for frontier exploration—sufficiently robust, entrepreneur-friendly, and globally competitive?
05
Meta’s acquisition of Manus is far more than a business transaction.
It is a microcosm—a concentrated reflection of the complex interplay among technology, capital, geopolitics, and individual ambition in our era.
For China’s AI industry, it serves as both a sobering wake-up call—exposing gaps and vulnerabilities in infrastructure and ecosystem—and a catalyst for deeper reflection: How can we systematically improve the soil of innovation, defend our developmental space amid global rule-setting contests, and truly convert talent and technological advantages into enduring industrial strength?
The future won’t be decided by isolated breakthroughs, but by the resilience and sophistication of entire ecosystems.
The Manus story has ended.
But how China’s AI industry writes its next chapter—there’s still a long road ahead.
|
3
2