{"id":3568,"date":"2025-07-22T13:55:54","date_gmt":"2025-07-22T05:55:54","guid":{"rendered":"https:\/\/bccmedianews.com\/?p=3568"},"modified":"2025-07-22T13:55:58","modified_gmt":"2025-07-22T05:55:58","slug":"manus-from-chinas-ai-beacon-to-silent-exit-in-127-days","status":"publish","type":"post","link":"https:\/\/bccmedianews.com\/?p=3568","title":{"rendered":"Manus: From China\u2019s AI Beacon to Silent Exit in 127 Days"},"content":{"rendered":"\n

An internal beta invitation code once soared to 100,000 RMB (approx. USD 13,800 \/ KRW 18.9 million). Four months later, the official website was completely shut down to Chinese users. In 2025, Chinese AI star startup Manus staged a shocking \u201croller coaster\u201d rise and fall. It burst onto the scene as a disruptor with the title of \u201cworld\u2019s first general-purpose AI agent,\u201d attracting over 2 million people to its waitlist in just three days, topping global benchmarks, and earning the title of \u201cChina\u2019s AI beacon.\u201d But just 127 days later, it exited quietly, with its core team hastily relocating to Singapore, leaving behind unanswered questions and a trail of chaos.
Did the fall of Manus expose the illusion of general-purpose AI agents? Or does it reveal deeper structural issues within the industry?<\/p>\n\n\n\n

From a 100,000 RMB Invite to Shutdown: The Fall of a Chinese AI Star<\/strong><\/p>\n\n\n\n

On March 6, 2025, at dawn, a demo video lit up China\u2019s tech community: on the left side of the screen, commands were entered into a chat box; on the right, real-time actions were executed. A so-called AI agent named Manus<\/em> completed a full-cycle task\u2014from price comparison on e-commerce platforms to generating an investment report\u2014within 15 minutes. The \u201cworld\u2019s first general-purpose AI Agent\u201d had arrived.<\/p>\n\n\n\n

Within three days, the waitlist on its website exceeded 2 million users, and the servers crashed due to overload. On China\u2019s secondhand market platform Xianyu, beta invite codes were being sold for as high as 100,000 RMB (approx. USD 13,800 \/ KRW 18.9 million). In its launch week, it even overtook OpenAI on the global GAIA benchmark leaderboard.<\/p>\n\n\n\n

However, by July, just four months later, the Manus website became inaccessible to Chinese IP addresses, and its social media accounts were wiped clean. Out of a 120-person team, only about 40 moved to Singapore, with the rest laid off. From peak to collapse\u2014only 127 days.<\/p>\n\n\n\n

Capital Frenzy: The 100,000 RMB Invite and the Myth of Chinese AI<\/strong><\/p>\n\n\n\n

In early 2025, China\u2019s AI sector was booming. DeepSeek had just proven the strength of Chinese foundational models, and the AI field was in a \u201chundred-model war.\u201d The market was hungry for the next disruptor.<\/p>\n\n\n\n

Manus satisfied that hunger on March 6. Unlike most AI products, it claimed to have an \u201cautonomous decision-making closed-loop system\u201d\u2014seen as a critical leap from \u201cassisted decision-making\u201d to \u201cautonomous execution.\u201d<\/p>\n\n\n\n

Its defining feature was the ability to execute tasks independently through a multi-agent architecture\u2014planner, executor, and verifier agents working together to handle everything from task decomposition and tool invocation to results validation. In a New York real estate example, Manus could break down user needs, call property APIs, write budget calculation programs, and output an interactive home-buying report.<\/p>\n\n\n\n

This power stemmed from the Monica<\/em> team\u2019s deep application of scaling laws: when model size (175B-level GPT-4o), training data (tens of millions of user logs), and compute (distributed cloud infrastructure) form a golden triangle, AI exhibits emergent behaviors in tool use and system planning.<\/p>\n\n\n\n

Even more impressive was its asynchronous execution engine. When prompted with \u201canalyze the past three years of Tesla stock volatility,\u201d Manus would spin up a cloud VM, autonomously complete data crawling, cross-checking, code writing, visualization, and more\u201417 steps\u2014without human intervention.<\/p>\n\n\n\n

This \u201cdigital intern\u201d model redefined productivity. GAIA benchmark results showed that Manus outperformed entry-level human analysts by 9.3% in complex problem-solving accuracy.<\/p>\n\n\n\n

In the demo, Manus resembled a true digital employee: receiving commands, decomposing steps, invoking tools, and delivering complete results. This harmony between \u201cknowing\u201d and \u201cdoing\u201d drove tech enthusiasts wild. Media dubbed it \u201cChina\u2019s AI beacon,\u201d and state broadcaster CCTV called it a \u201cnew paradigm of human-machine collaboration.\u201d<\/p>\n\n\n\n

But Manus\u2019 explosive popularity wasn\u2019t just about technical innovation. Its invite-only model drove extreme scarcity. Invite codes reached prices where people joked \u201cyou could hire a real assistant for six months instead.\u201d Amid the craze, co-founder Zhang Tao had to apologize publicly, citing limited server capacity and assuring waitlist users that expansion was underway.<\/p>\n\n\n\n

Capital followed closely. In April, Silicon Valley VC giant Benchmark led a $75 million Series B round, valuing Manus at $500 million\u2014setting a record for a Chinese AI application-layer startup.<\/p>\n\n\n\n

At the funding press event, founder Xiao Hong declared, \u201cManus will redefine human-machine collaboration, evolving from \u2018AI assistant\u2019 to \u2018AI colleague.\u2019\u201d At the time, no one questioned the vision.<\/p>\n\n\n\n

Yet hidden risks were already surfacing. Even as invitation prices climbed into six-figure territory, actual users reported issues: high task failure rates, infinite loops with complex commands, and hallucinations.<\/p>\n\n\n\n

Collapse: From Tech Disputes to Capital Pressure<\/strong><\/p>\n\n\n\n

The downfall began when Manus opened public registration. On May 12, it launched a three-tiered subscription model: Basic ($19\/month), Plus ($39\/month), and Pro ($199\/month). These prices were far beyond what Chinese consumers could accept\u2014especially when ChatGPT Plus cost only $20\/month.<\/p>\n\n\n\n

The market reaction was brutal. Monthly active users fell from a March peak of 20 million to 10 million by May. Paid conversion was under 3%, far below the industry average of 8\u201310%. Enterprise subscription renewal was only 45%.<\/p>\n\n\n\n

Worse problems surfaced. Tech insiders revealed that Manus mainly relied on Anthropic\u2019s Claude and Alibaba\u2019s Tongyi Qianwen\u2014making it a case of \u201cassembly-style innovation.\u201d<\/p>\n\n\n\n

Associate Professor Tan Jian of Beijing University of Posts and Telecommunications commented: \u201cIts \u2018autonomous execution\u2019 is essentially automated workflows built via API calls.\u201d<\/p>\n\n\n\n

User tests exposed fatal flaws. In March, only 4 of 8 tasks succeeded (50%). Its travel itinerary feature recommended nonexistent deals; its code often had redundancy.<\/p>\n\n\n\n

The real turning point came from capital regulation. Under the U.S. Outbound Investment Security Program<\/em>, American investments in China\u2019s critical tech\u2014including AI\u2014are restricted. To avoid scrutiny, Benchmark forced Manus to move its HQ to Singapore.<\/p>\n\n\n\n

That decision triggered a chain reaction: two-thirds of Manus\u2019 China staff were laid off; its partnership with Alibaba\u2019s Tongyi Qianwen ended; Chinese-language development stalled. By June, Manus ads began appearing in Singapore subway stations, while its Chinese social accounts went silent.<\/p>\n\n\n\n

\u201cIt\u2019s a textbook case of Silicon Valley capital sidestepping U.S. regulation,\u201d one industry insider noted. \u201cJust like how AI video startup HeyGen moved from Shenzhen to Los Angeles after raising funds.\u201d<\/p>\n\n\n\n

Double Squeeze: Compute Bottlenecks and a Broken Business Model<\/strong><\/p>\n\n\n\n

Moving to Singapore didn\u2019t save Manus\u2014it accelerated the collapse.<\/p>\n\n\n\n

U.S. export controls on GPUs to China meant Manus couldn\u2019t access NVIDIA\u2019s latest H100 chips. Insiders said its ability to handle complex tasks dropped by 30% compared to launch.<\/p>\n\n\n\n

Cost structure was another fatal flaw. As a multi-model AI agent, Manus cost about $2 per task. Its \u201cdiseconomies of scale\u201d were clear: more users meant bigger losses.<\/p>\n\n\n\n

Its flawed business model became more obvious in contrast. Competitor Genspark<\/em>, focused on web page analysis, reached $36 million in ARR (annual recurring revenue) in 45 days. Manus\u2019 \u201cdo-everything\u201d approach looked clumsy.<\/p>\n\n\n\n

\u201cConsumer-facing AI agents are money pits unless they find a killer vertical use case,\u201d one investor bluntly said.<\/p>\n\n\n\n

On July 19, co-founder Ji Yichao published a technical retrospective, indirectly admitting strategic mistakes: \u201cManus bet on context engineering instead of foundational models.\u201d<\/p>\n\n\n\n

He explained that during his last venture, their self-trained model became obsolete overnight.<\/p>\n\n\n\n

But it was too late. In just four months, the $75 million round was nearly burned through. Its technical foundation collapsed amid turmoil, and user trust in China hit zero.<\/p>\n\n\n\n

After the General AI Dream: Where Should Chinese AI Go Next?<\/strong><\/p>\n\n\n\n

The collapse of Manus pierced the bubble of general-purpose AI agents.<\/p>\n\n\n\n

Fang Han, CEO of Kunlun Wanwei, pointedly remarked: \u201cGeneral-purpose often means lack of specialization.\u201d<\/p>\n\n\n\n

Industry data shows a new path: In 2025, China’s core AI industry is projected to reach 1.1 trillion RMB (approx. USD 152 billion \/ KRW 208 trillion), with 65% from vertical markets.<\/p>\n\n\n\n

Enterprise AI applications have a 32.6% paid adoption rate\u2014far higher than the consumer market\u2019s 15.8%. Vertical large models in healthcare and finance now reach 93.6% accuracy, approaching human expert levels.<\/p>\n\n\n\n

Success stories are already emerging.<\/p>\n\n\n\n