04 May 2026
IP26063 | Global China, China-Shedding and Southeast Asia’s Regulatory Challenge
KEY TAKEAWAYS
• Beijing’s recent demand that Meta’s purchase of Manus, the Singapore-based Chinese AI start-up, be reversed shows the challenges of regulating Chinese-origin capital, talent, intellectual property and technology once they go offshore.
• For Southeast Asia, especially Singapore, opportunities to attract Al firms and investment sit alongside regulatory challenges and geopolitical pressure.
COMMENTARY
On 27 April, Beijing’s top state planner, the National Development and Reform Commission, issued a notice requiring that Meta reverse its US$2 billion acquisition of the AI startup Manus, citing national security concerns. Chinese regulators ordered the deal “unwound” to prevent the transfer of home-grown AI technology and talent to a US competitor. Meta made a commercial law argument, stating that the transaction “complied fully with applicable law” and that it anticipated “an appropriate resolution”. Beijing, on the other hand, applied a national security argument: in frontier AI, legality is not just about paperwork, it is about whether the transaction transfers strategic capability.
In January, after Meta had completed its acquisition of Manus, China’s commerce ministry announced an investigation into the sale. Two months later, Beijing warned the founders of AI firms against talent transfer, signalling that it treats AI capability as a non-portable asset. MicoMind, another Chinese AI start-up, has since distanced itself from US businesses after such warnings.
Beijing’s decision to block the Meta-Manus deal highlights a fundamental tension in its global strategy: while China encourages firms to expand internationally, it remains determined to anchor strategic AI talent and “dual-use” intellectual property within its own 15th Five-Year Plan ecosystem to ensure technological self-reliance.
Beijing’s Rationale for the Manus Ban
Beijing’s rationale for the ban is that Manus was not just a company being sold; it was a Chinese-origin AI capability potentially moving into a US Big Tech platform. The first concern is provenance. Manus presented itself as Singapore-based, but Chinese regulators appear to be asking deeper questions: where did the early R&D take place, who built the product, what data and infrastructure supported it, and whether the transaction would shift control out of Chinese hands and allow Chinese-origin AI know-how to be absorbed into Meta’s ecosystem.
The second concern is control. In an AI firm, control is not just about who owns the shares; it is about who controls the technical core and the way the technology is integrated into a larger platform. If Meta absorbed Manus, Beijing could view that as the transfer of strategic capability, not just a commercial acquisition.
The third concern is talent. In frontier AI, people are often part of intellectual property. Founders, scientists and engineers are carriers of knowledge transfer.
The fourth concern is regulatory-arbitrage risk. Companies may internationalise for legitimate reasons – to attract customers, gain talent, increase capital, diversify risk or comply with foreign regulations. Beijing’s narrower concern is that in strategic technologies, a transaction can make the China connection less visible commercially – through ownership, branding, customer relationships or integration into a foreign platform – while the asset’s core value may still depend on Chinese-origin AI talent, R&D, models, data, infrastructure or know-how. This is where Beijing views the Manus sale as a national security concern.
China-Origin AI Firms as Strategic National Assets
To be sure, not all cases of Chinese firms moving or establishing headquarters overseas are the same. In early 2026, TikTok and its parent company ByteDance successfully established a majority American-owned joint venture, TikTok USDS Joint Venture LLC, to avoid a US ban on the popular video-sharing app. Shein, the fashion e-commerce platform, moved its headquarters to Singapore in 2022. Temu, another e-commerce platform, lists Dublin as its principal office. Both Shein and Temu run on Chinese supply chains, merchants, logistics and consumer data.
But unlike e-commerce firms, the sensitive asset for AI firms like Manus and MicoMind is not a finished product but research capacity itself, which includes scientists and engineers, model know-how, access to computing resources, and intellectual property.
Singapore and ASEAN Exposure
For the ASEAN countries, especially Singapore, opportunities to attract Al firms and investment pose regulatory challenges and geopolitical pressure. Singapore’s various strengths make it attractive to such firms: it is legally predictable, English-speaking, globally connected, credible to Asian and Western investors, and increasingly visible as an AI hub. For Chinese-origin AI firms trying to appear global rather than China-based, Singapore offers neutrality and legitimacy. But neutrality is not insulation for the island-state as Singapore firms can become unwitting conduits for Chinese firms to access restricted chips, cloud services, models, data, or intellectual property from the United States and other Western countries.
The risk is uneven across ASEAN. In Singapore, the issue is whether a firm with headquarters in the country, its capital structures, and the existence of compliance teams reflect real independence. In Malaysia, the concern may be centred around the growing number of data centres established by Chinese firms, their cloud infrastructure, access to computing resources, and chip procurement. In Vietnam, the establishment of manufacturing plants by Chinese firms may mask their reliance on China-based inputs. In Thailand and Indonesia, Chinese-origin platforms may depend on China-linked logistics, fintech, merchants, and data systems. In the Philippines, alliance politics make foreign-controlled digital systems more sensitive because data access, disinformation, cybersecurity, and infrastructure dependence can acquire security significance.
Policy Response
Southeast Asia should adopt a trusted investment checklist for AI and digital firms. Such a checklist should ask who ultimately owns, controls, finances, and can direct the firm; whether founders, investors or board members retain veto power or informal influence; how the firm’s core technology was created, moved and controlled; how chips and cloud services are procured; where data flows; whether cybersecurity controls are auditable; what export-control obligations may be triggered; and whether personnel, assets or affiliates remain subject to foreign legal or reporting duties. This should apply to any firm whose structure obscures strategic control or sensitive technology transfer, not only Chinese-origin firms.
ASEAN should use the ASEAN Working Group on AI Governance, chaired by Singapore’s Infocomm Media Development Authority (IMDA), to develop a Trusted AI Investment Protocol as a companion to the ASEAN AI governance framework. The Working Group already has a mandate to shape regional AI frameworks, standards, safety discussions, regulation, industry assessments, implementation roadmaps, research and capacity building. The Expanded ASEAN Guide on AI Governance and Ethics – Generative AI is a useful foundation for responsible AI, but China-shedding raises harder questions about investment screening, compute governance, export-control compliance, and strategic technology transfer. Common expectations would reduce regulatory arbitrage and prevent an AI race to the bottom.
Conclusion
As Chinese capital goes abroad, there will be unintended consequences for neutral financial hubs like Singapore. Washington will not let Beijing write the rules for global AI consolidation. At the same time, Beijing has demonstrated that it is prepared to leverage its foreign investment security review mechanism to publicly warn its AI startups that are considering the same path, that is, offshore headquarters, US capital, global customers and exit from China when they join the Big Tech league.
The case of Manus shows the challenges of regulating Chinese-origin capital, talent, intellectual property and technology once they go offshore. The policy implication, as far as Beijing is concerned, is that it is prepared to securitise offshore AI firms even if they are compliant from a commercial standpoint.
Looking ahead, Southeast Asia needs regulation that is open but not naive, and security-conscious but not identity- or geopolitically driven. It should welcome real innovation that generates tangible domestic value, such as local R&D activity and talent development, while strengthening regulatory controls and scrutiny of investments and firms dealing with sensitive technologies, including AI.
Stefanie Kam is an Assistant Professor in the China Programme at the Institute of Defence and Strategic Studies (IDSS), S. Rajaratnam School of International Studies (RSIS).
KEY TAKEAWAYS
• Beijing’s recent demand that Meta’s purchase of Manus, the Singapore-based Chinese AI start-up, be reversed shows the challenges of regulating Chinese-origin capital, talent, intellectual property and technology once they go offshore.
• For Southeast Asia, especially Singapore, opportunities to attract Al firms and investment sit alongside regulatory challenges and geopolitical pressure.
COMMENTARY
On 27 April, Beijing’s top state planner, the National Development and Reform Commission, issued a notice requiring that Meta reverse its US$2 billion acquisition of the AI startup Manus, citing national security concerns. Chinese regulators ordered the deal “unwound” to prevent the transfer of home-grown AI technology and talent to a US competitor. Meta made a commercial law argument, stating that the transaction “complied fully with applicable law” and that it anticipated “an appropriate resolution”. Beijing, on the other hand, applied a national security argument: in frontier AI, legality is not just about paperwork, it is about whether the transaction transfers strategic capability.
In January, after Meta had completed its acquisition of Manus, China’s commerce ministry announced an investigation into the sale. Two months later, Beijing warned the founders of AI firms against talent transfer, signalling that it treats AI capability as a non-portable asset. MicoMind, another Chinese AI start-up, has since distanced itself from US businesses after such warnings.
Beijing’s decision to block the Meta-Manus deal highlights a fundamental tension in its global strategy: while China encourages firms to expand internationally, it remains determined to anchor strategic AI talent and “dual-use” intellectual property within its own 15th Five-Year Plan ecosystem to ensure technological self-reliance.
Beijing’s Rationale for the Manus Ban
Beijing’s rationale for the ban is that Manus was not just a company being sold; it was a Chinese-origin AI capability potentially moving into a US Big Tech platform. The first concern is provenance. Manus presented itself as Singapore-based, but Chinese regulators appear to be asking deeper questions: where did the early R&D take place, who built the product, what data and infrastructure supported it, and whether the transaction would shift control out of Chinese hands and allow Chinese-origin AI know-how to be absorbed into Meta’s ecosystem.
The second concern is control. In an AI firm, control is not just about who owns the shares; it is about who controls the technical core and the way the technology is integrated into a larger platform. If Meta absorbed Manus, Beijing could view that as the transfer of strategic capability, not just a commercial acquisition.
The third concern is talent. In frontier AI, people are often part of intellectual property. Founders, scientists and engineers are carriers of knowledge transfer.
The fourth concern is regulatory-arbitrage risk. Companies may internationalise for legitimate reasons – to attract customers, gain talent, increase capital, diversify risk or comply with foreign regulations. Beijing’s narrower concern is that in strategic technologies, a transaction can make the China connection less visible commercially – through ownership, branding, customer relationships or integration into a foreign platform – while the asset’s core value may still depend on Chinese-origin AI talent, R&D, models, data, infrastructure or know-how. This is where Beijing views the Manus sale as a national security concern.
China-Origin AI Firms as Strategic National Assets
To be sure, not all cases of Chinese firms moving or establishing headquarters overseas are the same. In early 2026, TikTok and its parent company ByteDance successfully established a majority American-owned joint venture, TikTok USDS Joint Venture LLC, to avoid a US ban on the popular video-sharing app. Shein, the fashion e-commerce platform, moved its headquarters to Singapore in 2022. Temu, another e-commerce platform, lists Dublin as its principal office. Both Shein and Temu run on Chinese supply chains, merchants, logistics and consumer data.
But unlike e-commerce firms, the sensitive asset for AI firms like Manus and MicoMind is not a finished product but research capacity itself, which includes scientists and engineers, model know-how, access to computing resources, and intellectual property.
Singapore and ASEAN Exposure
For the ASEAN countries, especially Singapore, opportunities to attract Al firms and investment pose regulatory challenges and geopolitical pressure. Singapore’s various strengths make it attractive to such firms: it is legally predictable, English-speaking, globally connected, credible to Asian and Western investors, and increasingly visible as an AI hub. For Chinese-origin AI firms trying to appear global rather than China-based, Singapore offers neutrality and legitimacy. But neutrality is not insulation for the island-state as Singapore firms can become unwitting conduits for Chinese firms to access restricted chips, cloud services, models, data, or intellectual property from the United States and other Western countries.
The risk is uneven across ASEAN. In Singapore, the issue is whether a firm with headquarters in the country, its capital structures, and the existence of compliance teams reflect real independence. In Malaysia, the concern may be centred around the growing number of data centres established by Chinese firms, their cloud infrastructure, access to computing resources, and chip procurement. In Vietnam, the establishment of manufacturing plants by Chinese firms may mask their reliance on China-based inputs. In Thailand and Indonesia, Chinese-origin platforms may depend on China-linked logistics, fintech, merchants, and data systems. In the Philippines, alliance politics make foreign-controlled digital systems more sensitive because data access, disinformation, cybersecurity, and infrastructure dependence can acquire security significance.
Policy Response
Southeast Asia should adopt a trusted investment checklist for AI and digital firms. Such a checklist should ask who ultimately owns, controls, finances, and can direct the firm; whether founders, investors or board members retain veto power or informal influence; how the firm’s core technology was created, moved and controlled; how chips and cloud services are procured; where data flows; whether cybersecurity controls are auditable; what export-control obligations may be triggered; and whether personnel, assets or affiliates remain subject to foreign legal or reporting duties. This should apply to any firm whose structure obscures strategic control or sensitive technology transfer, not only Chinese-origin firms.
ASEAN should use the ASEAN Working Group on AI Governance, chaired by Singapore’s Infocomm Media Development Authority (IMDA), to develop a Trusted AI Investment Protocol as a companion to the ASEAN AI governance framework. The Working Group already has a mandate to shape regional AI frameworks, standards, safety discussions, regulation, industry assessments, implementation roadmaps, research and capacity building. The Expanded ASEAN Guide on AI Governance and Ethics – Generative AI is a useful foundation for responsible AI, but China-shedding raises harder questions about investment screening, compute governance, export-control compliance, and strategic technology transfer. Common expectations would reduce regulatory arbitrage and prevent an AI race to the bottom.
Conclusion
As Chinese capital goes abroad, there will be unintended consequences for neutral financial hubs like Singapore. Washington will not let Beijing write the rules for global AI consolidation. At the same time, Beijing has demonstrated that it is prepared to leverage its foreign investment security review mechanism to publicly warn its AI startups that are considering the same path, that is, offshore headquarters, US capital, global customers and exit from China when they join the Big Tech league.
The case of Manus shows the challenges of regulating Chinese-origin capital, talent, intellectual property and technology once they go offshore. The policy implication, as far as Beijing is concerned, is that it is prepared to securitise offshore AI firms even if they are compliant from a commercial standpoint.
Looking ahead, Southeast Asia needs regulation that is open but not naive, and security-conscious but not identity- or geopolitically driven. It should welcome real innovation that generates tangible domestic value, such as local R&D activity and talent development, while strengthening regulatory controls and scrutiny of investments and firms dealing with sensitive technologies, including AI.
Stefanie Kam is an Assistant Professor in the China Programme at the Institute of Defence and Strategic Studies (IDSS), S. Rajaratnam School of International Studies (RSIS).
