On March 25, it was Manus co-founders Xiao Hong and Ji Yichao prohibited from leaving China after a meeting with the National Development and Reform Commission in Beijing after the Ministry of Commerce launched a national security review of their company’s $2 billion purchase of Meta.
By April 2, Beijing’s position articulated a explicit message: it supports cross-border operations, but only those that “comply with Chinese laws and regulations and follow proper procedure.”
The Manus case is the clearest signal that, amid great power competition, the strategy of “washing Singapore” has reached a critical limit. This has implications not only for up-and-coming Chinese tech firms, but also for smaller economies like Singapore that seek to bridge the increasingly fragmented global AI ecosystem by hosting these firms.
Duel pressures
The strategy that Manus chose is commonly known as “Bathing in Singapore.” For some time, Chinese companies seeking international expansion and fewer domestic regulatory hurdles have moved headquarters or operations to more open Southeast Asian economies, particularly Singapore.
From there, Chinese tech firms can more easily access capital and international customers. Manus, originally founded in China, got this strategy in an extreme. In 2025, the AI agent startup made a sudden decision to relocate entirely to Singapore, laying off most of its China-based employees and shutting down domestic operations.
The move was seen by management as essential in making the company “clean” for acquisition by US tech giant Meta. But in trying to please Washington, Manus appears to have crossed a red line in Beijing.
The pressure from the United States and China on Chinese-founded AI firms is now unmistakable. In recent years, Washington has pursued active control of Chinese AI companies, framing the AI race as a matter of national security.
While most of the legal remedies target firms operating within China or companies controlled by China, the pressure has also extended to Singapore and Malaysia. This was evident in the US enforcement actions in 2025 and 2026 targeting entities related to those countries for allegedly diverting limited Nvidia chips to China.
Washington too under pressure Malaysia to introduce chip export licensing and warned against deploying Huawei Ascend chips. These steps signal clearly and firmly that the US does not want Singapore and Malaysia to become easy paths for Chinese firms to circumvent restrictions.
From Beijing, pressure operates on a different logic, but leads to a similar result. The Ministry of Commerce doctrine articulated through the Manus review establishes a principle: technological nationality follows where the technology is developed, not where the company is registered.
What matters to Beijing is where the model was trained, where the data was generated, and where the engineers who built it got their expertise.
China does not object to companies moving their headquarters, as long as the core technology, data and intellectual property are not transferred. However, the Manus case could set a dangerous precedent for future Chinese AI startups that build capacity domestically at a lower price and later sell those assets to large Western tech firms.
ByteDance, on the other hand, exemplifies what seems acceptable to Beijing. An early adopter of the “wash Singapore” strategy, it achieved international success through TikTok and is expanding AI capacity in Malaysia to bypass US chip export controls.
While ByteDance has also faced intense US scrutiny, Beijing has largely tolerated its approach because it has not cut ties with China. Its core operations, data and talent remain within the Chinese tech ecosystem. For ByteDance, Southeast Asia serves as an extension rather than an escape.
Beijing appears willing to allow companies to use Singapore as a launch pad, but not as an exit ramp.
Bridge economy link
By hosting Chinese firms, Singapore and Malaysia have positioned themselves as bridges in a dual global AI landscape. Through foreign direct investment, talent inflows, and revenue-generating infrastructure such as data centers, this connectivity model provides significant economic benefits to these host countries.
However, the Manus case and parallel US enforcement actions expose structural weaknesses in this model. It was built on strong legal frameworks and investment-friendly ecosystems, but legal complexities around technology and data transfer have intensified as major powers such as the US and China exert more and more pressure.
Countries like Singapore cannot afford to be seen as conduits for evading sanctions or facilitating the transfer of Chinese intellectual property to American buyers.
The implication is that countries like Singapore may need to strengthen the regulatory framework to meet compliance expectations from both powers. In practice, this means developing more rigorous review mechanisms for cross-border AI transactions, ensuring that companies comply with US export controls and Chinese regulations on data and IP transfers.
Some countries have responded to the dual landscape of AI by following a completely different path. The Gulf states, especially the United Arab Emirates and Saudi Arabia, have invested heavily in domestic computing capacity and sovereign AI models, apparently hoping to turn the current bipolar AI system into a multipolar one.
South Korea, Japan and India are pursuing similar strategies. However, sovereign AI is costly, requiring energy, computing power and talent that many countries lack. For most Southeast Asian economies, the bridge model remains the most sustainable short-term strategy, making it critical to strike the right regulatory balance.
Neutrality as proof of work
While the outcome of the Manus case remains uncertain, one trend is clear: the AI era will not follow previous patterns of offshoring, where companies freely used offshore structures to maximize capital gains.
States are tightening their control over intellectual property, talent and data. The era in which a small city-state or offshore holding structure can act as frictionless infrastructure between rival technology ecosystems is ending.
In an era of rising technological nationalism, each country in the AI supply chain must chart its own course – and neutrality is no longer an attitude, but a proof of work.
Minghao Sun is involved in AI strategy in the UAE, has worked as a consultant to Gulf ministries and advised government entities on technology policy, economic policy and geopolitics with Whiteshield Advisors. He has a degree in sociology from the University of Chicago.
