Manus Exits China Market Amid $500 Million Valuation, Prompting Concerns Over AI Agent Sector Outlook

Manus, once regarded as a rising star in China’s AI Agent field, has abruptly withdrawn from the Chinese market despite achieving a valuation of US$500 million. The company has initiated a large-scale layoff, deactivated domestic accounts, and restricted access from Chinese IP addresses. Reports indicate that Manus plans to relocate its headquarters from China to Singapore, accompanied by a significant downsizing of its domestic workforce. Of the original 120 employees, only around 40 core technical staff will transfer to the Singapore office, with the remainder laid off. Severance compensation is reportedly generous, offering terms such as N+3 or 2N.

Manus issued a public response stating the decision was made based on operational efficiency, and it would continue to focus on core business development to improve overall performance. In regard to reports of its “departure,” a source quoted by Huxiu stated the company is functioning normally and denied a relocation of headquarters, clarifying that Manus was launched as an international product in March. The recent steps, the source explained, aim to ensure compliance with international standards while serving global users.

Manus positions itself as a general-purpose AI Agent capable of performing complex tasks through web browsing and tool operation. The system displays real-time step-by-step execution via a virtual machine interface, seeking to bridge the gap between ideation and action. Unlike conventional chatbots that merely provide responses, Manus integrates AI models with automation systems, reducing the need for manual intervention. Following the launch of DeepSeek, it became another prominent example of Chinese innovation in AI.

Upon its release on March 6, the platform quickly gained traction. Derived from the Latin term “Mens et Manus,” meaning “Mind and Hand,” Manus amassed a waitlist of over two million users within three days. Due to its invite-only beta model, access codes were resold for as much as RMB 100,000. Just five days post-launch, Manus announced a strategic partnership with Alibaba’s Tongyi Qianwen team. The collaboration aimed to implement the full Manus functionality using the Tongyi Qianwen open-source models and domestic computing platforms to expand its footprint in China. However, the process was suspended following the company’s market withdrawal.

In late April, Manus completed a US$75 million Series B funding round led by Benchmark, elevating its valuation to approximately US$500 million—representing a fivefold increase. The company stated that proceeds from the round would support expansion across the U.S., Japan, and Middle East markets. On May 12, Manus opened registration to all users, and over one million signed up on the first day.

The decision to exit China was not entirely unforeseen. According to the South China Morning Post, new U.S. regulations effective from January require capital and high-end technologies invested in overseas AI and related fields to undergo security review. Consequently, the Benchmark-led investment was subject to scrutiny by the U.S. Treasury, and investors reportedly requested that Manus relocate its headquarters to avoid regulatory risks and integrate into the international ecosystem.

However, attributing the move solely to investment-related pressure may be an oversimplification. The company’s Singapore entity, “Butterfly Effect,” was registered as early as August 2023 and is wholly owned by a Cayman Islands entity, suggesting that the founding team had long envisioned a global development strategy for Manus.

According to Zhidao, two primary factors may have driven the decision. First, limitations in computing power have significantly hindered development. Due to GPU export restrictions, Manus was unable to secure the necessary high-end resources to support iterative development. As a platform that requires simultaneous deployment of multiple large models, lack of access to advanced GPUs posed a serious threat to competitiveness. Relocation to Singapore would allow Manus to tap into a more open computing supply chain and bypass technical constraints.

Second, the company encountered commercial challenges. During its initial beta in March, monthly active users reached 20 million. However, after public registration opened in May, this number dropped sharply to about 10 million, with user retention decreasing by half. Manus introduced subscription plans priced at US$19, US$39, and US$199 per month, but many users viewed the pricing as too high, and the lack of functional differentiation limited effectiveness.

Zhidao noted that these cumulative issues narrowed Manus’s operating space in China. As the AI Agent industry transitions from conceptualization to real-world validation, the inability to generate sufficient Annual Recurring Revenue (ARR) to match high valuations risks rapid withdrawal of both capital and market support.

Founder Xiao Hong shared sentiments on social media, stating that sacrificing short-term revenue and adjusting business models for broader accessibility was a worthwhile decision. He emphasized that ARR may not be the correct incentive function for building globally competitive products, especially given the challenges that extend beyond user value and operations. He added that the journey itself offered valuable growth for the team and hoped that successful outcomes would validate the ability of China-born founders to create global products.

While Manus's initial reception in China was strong, its exit is expected to have limited impact on the broader domestic AI Agent sector. A recent Goldman Sachs report highlighted the enterprise-driven nature of China's AI landscape, where many state-owned enterprises and government entities are adopting localized models, such as DeepSeek. Manus’s focus on consumer and developer markets diverged from this mainstream direction. Goldman Sachs also noted that the conversion rate for consumer AI tools remains low.

Therefore, Manus’s withdrawal reflects more of a response to international geopolitical dynamics than a failure in business model or market demand. The domestic market is increasingly led by enterprise-focused AI applications and privatized models for government and industry. Sectors such as finance and manufacturing are actively implementing tailored agent tools, and domestic players continue to expand within these verticals. As a result, Manus’s departure may even clear the way for other firms, prompting the industry to shift focus from speculative concepts to productivity-driven outcomes.

Despite its move to Singapore, Manus continues to face competitive challenges. U.S. regulatory approval for Benchmark’s investment remains pending, and Singapore’s high labor costs will further pressure operations. The company must now demonstrate that its new team structure can support rapid product development and gain market recognition.

Globally, several startups and tech giants are also investing in the AI Agent space. Genspark, founded by a former Baidu executive, has already exceeded US$36 million in short-term revenue, while Manus’s ARR remains under US$10 million, with declining traffic trends.

Zhidao concluded that AI startups must strike a delicate balance among investor expectations, technical constraints, and real-world demand. In an environment of increasing regulatory scrutiny, high model costs, and intensified competition, Chinese founders aspiring to globalize must develop the resilience to maintain steady operations, endure external pressures, and navigate market cycles. Whether Manus can secure a new foothold in Singapore and sustain its technological lead remains to be seen.