The Chinese threat to Singapore

As Air Force One touched down in Beijing this week for the first US presidential visit to China in nearly a decade, the world’s attention has understandably fixed on the more obvious flashpoints: Iran’s closure of the Strait of Hormuz, tariffs, the future of Taiwan. Yet among the quieter items on the agenda is the establishment of a bilateral AI safety dialogue, which may prove the most consequential of all. For in the Strait of Malacca, arguably the most consequential geostrategic contest of the 21st century is already playing out. China’s decision last month to block US tech giant Meta’s acquisition of Manus, an artificial intelligence (AI) platform originally developed in China, is not merely the story of a failed deal. It marks a decisive shift in the AI race from commercial to security rivalry, and holds dire consequences for one of the most successful small-state strategies of the modern era.

Singapore’s foreign policy is a Thucydidean balancing act par excellence. Limited by size and resource scarcity, the Little Red Dot has leveraged its historical inheritance as a colonial entrepôt to great effect, famously going ‘from third world to first’ in 30 years through free-market capitalism and the export of labour-intensive manufactures (in particular, electronics) based on comparative advantage, in the process becoming indispensable to global supply chains. Moreover, in identifying emerging industries and providing both the legal confidence and infrastructural capacity required by foreign firms to invest, Singapore, perhaps more than any other ‘middle power’, has refined the art of geostrategic horizon scanning. In the realm of AI, its National AI Strategy 2.0 has already seen banking and tech conglomerates establish over 50 ‘AI Centres of Excellence’ across the city-state.

Yet as Fukuyama’s homily on globalisation rings hollow amid escalating US-China competition and the recognition that ‘supply chain security is national security’, Singapore finds itself straddling an increasingly untenable contradiction. It is caught between Washington’s determination to remain technologically superior to China and Beijing’s determination to upend that superiority to its own advantage. Now, having positioned itself as a global AI hub, Singapore as a middle power unsurprisingly finds itself caught in the middle.

Historically, Singapore has walked this tightrope gracefully. Early CIA meddling aside, since the offshoring of National Semiconductor in 1968, Singapore has experienced deepening linkages with the US, becoming a lodestone for American investment and lynchpin of US-centred supply chains in advanced technologies. Following the emergence of AI, there was a proliferation in the number of formal and informal partnerships aimed at nurturing collaborative AI ecosystems, including the US-Singapore Critical and Emerging Technology Dialogue and the Pax Silica Declaration. Simultaneously, as China has become a more active economic player under its ‘Going Out’ policy, Singapore has reacted positively by providing Chinese enterprises with the regulatory freedom, international capital flows, and ostensibly neutral branding necessary to globalise effectively, culminating in Chinese investments in Singapore outstripping those of the US last year.

Now, as both Washington and Beijing recognise the national security implications of AI and seek to weaponise technological access, Singapore must reconcile itself with the truism that no one can serve two masters. In April 2025, Butterfly Effect – a Beijing-origin AI start-up founded three years prior by millennial entrepreneur Xiao Hong – raised $75 million in a Series B funding round led by Silicon Valley venture capital firm, Benchmark. Almost immediately, the US Treasury launched an investigation under the auspices of the Outbound Investment Security Program (OISP), which had entered into force in January 2025 with the explicit aim of limiting US capital flows to Chinese firms operating in the AI sector. The company’s response was equally immediate: in July 2025, Butterfly Effect relocated its headquarters and 40 core staff to Singapore, shuttered its Chinese-language social media accounts, and shelved plans for a Chinese version of its agentic AI, Manus.

At the time, such extreme measures represented a broader phenomenon, referred to as ‘Singapore washing’, which has in recent years enabled Chinese firms to escape the mainland’s reputational straitjacket and access global capital by effectively rebranding as Singapore-based multinationals. Confronted by a storm of US investment restrictions and growing domestic competition from Chinese tech giants ByteDance and Baidu, Singapore’s perceived neutrality in this instance provided safe harbour to young start-up Butterfly Effect, whose AI product Manus fell foul of home market saturation and haemorrhaged 10 million active monthly users in May 2025. By maintaining its access to global capital through Singapore, the company achieved a $2 billion valuation within six months and in December 2025 announced it was selling Manus to Meta in a highly publicised deal, which was framed as accelerating the US tech giant’s agentic AI ambitions.

The deal, however, never went through. In January 2026, China’s Ministry of Commerce instead announced a review of Meta’s acquisition, and on 27 April issued a one-line order via the National Development and Reform Commission (NDRC), a Mao-era champion of the planned economy, prohibiting the deal outright and ‘requiring the parties involved to withdraw the acquisition transaction’. That the decision was reportedly made by China’s National Security Commission, a body chaired by General Secretary Xi Jinping, is a testament to the severity with which Beijing viewed the loss of Manus. Butterfly Effect’s founders were even barred from leaving the country following a summons from the NDRC.

Such actions reflect Beijing’s recognition that AI is no longer merely a commercial asset. In fact, since at least 2008, the country’s leadership has identified China’s ability to independently innovate in science and technology as critical to what it describes as ‘core national power’. At a macroeconomic level, China’s leaders assess that the world has entered a technological era characterised by ‘information and intelligentisation’, with potentially dual-use agentic AI such as Manus (capable of executing complex workflows autonomously) serving as the barycentre for both economic and military revolutions. Reducing dependency on US expertise by developing an ‘original source innovation’ ecosystem is thus imperative to China’s geostrategic ambitions. Yet, by its own admission, the country lacks highly skilled talent at scale. Losing what talent it does have to US tech giants is simply unacceptable.

For this reason, the strait through which Chinese AI firms have passed on their way to global markets looks set to narrow, and even close altogether. After all, ‘Singapore-washing is only credible and effective for companies that fully cut off their operational ties to China’. Such were Butterfly Effect’s intentions, and Beijing prevented this from occurring.

The immediate consequences for China in doing so are hardly cost-free. In reasserting its sovereignty, Beijing likely truncates growth both by restricting global funding and by forcing innovators back onto a domestic science and technology base stifled by limited capital, poor organisation and a lack of robust supply-chains. For Singapore, however, the implications are more structurally dire. Since the days of Lee Kuan Yew, the city’s development model has rested on regulatory neutrality as one of its keystones. In this instance, its value proposition was that, by being properly incorporated in Singapore, Butterfly Effect could access global capital and seize opportunities for growth overseas, especially in US markets. By extending the reach of its regulatory authority into Singapore, Beijing has systematically undermined such a proposition and almost certainly spooked other Chinese-origin firms into reconsidering the viability of Singapore as a vehicle for their AI ambitions.

This matters enormously for Singapore’s future FDI flows. Singapore’s model has always been acutely exposed to shifts in comparative advantage, given its small domestic market and extreme dependence on global trade. Singapore’s approach of anticipating promising sectors and showering foreign firms with incentives works only as long as the latter see the city-state as a stable, credible, and legally sovereign platform; once that credibility is in question, the incentives on offer become secondary.

Here lies a wider lesson, and one that extends beyond Singapore to encompass all small and middle powers. Over the past half-century, the Singapore story was built on a Fukuyaman promise that globalisation and the rules-based order would endure, and that a city-state with no hinterland could survive, grow, and prosper through economic and technological integration. Increasingly, however, great powers are asserting extraterritorial authority over technologies and companies they consider to be strategically essential. In the process, they are splintering global ecosystems and undermining the neutrality upon which smaller powers depend for their existence. Butterfly Effect and the struggle over Manus is a reminder that the space in which Singapore has operated is narrowing, and that the AI race, far from expanding it, may be closing it further.