Executive Summary
As of early 2026, a pronounced divergence has emerged in how investors across Chinese and American markets perceive, price, and respond to artificial intelligence developments. In the United States, rapid capability gains have triggered an ‘AI scare trade’—a systematic de-rating of software incumbents and wealth management firms on fears that AI will compress established profit pools. In China, the same developments are greeted with enthusiasm: investors are chasing newly listed AI pure plays, buoyed by a structurally insulated domestic market, competitive model benchmarks, and abundant private capital.
Singapore occupies a uniquely consequential position in this divergence. As a regional financial hub, home to strategically positioned investment firms, and a node through which global capital accesses Asian technology markets, Singapore is simultaneously a beneficiary of and a bridge between these two distinct investment regimes. This case study examines the structural drivers of the divergence, its key market manifestations, and the specific implications for Singapore’s financial ecosystem.
- Background and Context
1.1 The Global AI Investment Landscape
The commercialisation of large language models (LLMs) since late 2022 has restructured global equity markets around a new axis: which companies will gain value from AI, and which will be disrupted by it. By early 2026, this question is being answered differently depending on which market one examines.
In the United States, the dominant incumbents—enterprise software vendors, financial advisory platforms, and knowledge-work services—face genuine disruption risk as AI agents demonstrate the capacity to perform tasks previously reserved for skilled human labour. Investors have responded by selling these incumbents and concentrating capital in a narrow band of frontier AI developers and infrastructure providers.
China’s trajectory has been shaped by a different set of conditions: state industrial policy, competitive model development under resource constraints, and a domestic market effectively ring-fenced from Western AI providers. The emergence of DeepSeek in late 2024 and early 2025 crystallised the thesis that Chinese AI laboratories could produce frontier-capable models at dramatically lower cost, accelerating both domestic adoption and investor conviction.
1.2 Key Market Events (2025–2026)
Event Date Market Impact
DeepSeek R1 release demonstrates cost-efficient reasoning Jan 2025 Global semiconductor sell-off; Chinese AI re-rating begins
MiniMax Group IPO on Hong Kong Stock Exchange Jan 2026 Stock up 488% by February
Zhipu (Knowledge Atlas) IPO, Hong Kong Jan 2026 Stock up 524% by February
GLM-5 claims top global open-source benchmark (Artificial Analysis) Feb 2026 Zhipu rally extends; sector confidence rises
Anthropic raises $30B at $380B valuation Feb 2026 Halo effect lifts Chinese AI comparables
OpenAI approaches $850B+ valuation in new round Feb 2026 Reinforces AI premium valuations globally
ByteDance video-making app (Jianying) roll-out Feb 2026 Rally in China film and media sector stocks
- The United States: The ‘AI Scare Trade’
2.1 Mechanism and Logic
The term ‘AI scare trade’ describes a pattern of capital rotation in which investors sell companies with established, high-margin business models on the premise that AI will commoditise their core value proposition. The affected cohort is broad: it spans enterprise SaaS vendors, business process outsourcing firms, legal and accounting services platforms, and wealth management companies.
The underlying logic is rooted in profit pool theory. US technology incumbents have historically commanded premium valuations by erecting durable competitive moats—proprietary data, network effects, regulatory capture, and switching costs. AI, and in particular agentic AI systems capable of autonomous reasoning and task execution, threatens to erode these moats by lowering the marginal cost of intelligence-intensive work to near zero.
2.2 Sectoral Vulnerability
The sectors most exposed to the US scare trade include the following. First, enterprise software: tools built on proprietary workflow automation are susceptible to displacement by AI agents that can be instructed in natural language, reducing the value of structured proprietary interfaces. Second, financial advisory: robo-advisory and algorithmic wealth management platforms face existential competition from AI systems capable of personalised, dynamic portfolio construction at negligible cost. Third, knowledge work services: firms providing legal research, document review, market analysis, and compliance monitoring face direct automation risk.
Notably, semiconductor and infrastructure providers are broadly exempt from the scare trade. NVIDIA and its supply chain have benefited from the AI buildout, as have cloud infrastructure providers whose capacity underlies AI workloads.
2.3 Structural Constraints on US AI Champions
A further source of US investor anxiety is the concentration of value creation in a small number of unlisted entities. OpenAI and Anthropic—considered the frontier developers—are private companies. Public investors seeking pure-play LLM exposure in the United States have limited options, concentrating capital in adjacent infrastructure plays and amplifying the valuation divergence between incumbents (sold) and AI infrastructure (bought). - China: The AI Penetration Trade
3.1 Structural Insulation and Competitive Dynamics
China’s AI investment narrative is structurally distinct for reasons that are both regulatory and geopolitical. Foreign LLMs—including OpenAI’s GPT series and Anthropic’s Claude—face substantial barriers to deployment in the Chinese domestic market. Regulatory requirements around content moderation, data localisation, and algorithmic governance effectively preclude foreign AI providers from competing for the mass market. This creates an asymmetric competitive environment in which domestic developers enjoy captive addressable markets.
As Allspring Global Investments portfolio manager Gary Tan observed, foreign LLMs ‘have limited access to the domestic market, giving local model makers a clear run.’ This structural insulation removes the principal source of competitive anxiety that characterises the US market and redirects investor attention toward growth and penetration metrics rather than incumbent defence.
3.2 The Listed Pure-Play Premium
The scarcity of publicly listed LLM developers globally confers a structural premium on Chinese AI stocks. MiniMax and Zhipu both debuted on the Hong Kong Stock Exchange in January 2026 and rapidly appreciated to multiples that reflected investor willingness to pay for exposure to frontier AI development. By late February 2026, Zhipu had appreciated 524 percent and MiniMax 488 percent from their listing prices.
This premium is amplified by the halo effect of US private round valuations. Anthropic’s February 2026 fundraising at a $380 billion valuation and OpenAI’s approach to an $850 billion-plus valuation serve as reference points that encourage re-rating of Chinese comparables, even where the underlying business fundamentals differ substantially.
3.3 Benchmark Competitiveness and Cost Efficiency
Chinese AI laboratories have demonstrated competitive performance at cost structures significantly below those of US frontier developers. DeepSeek’s R1 and R2 model series established that capable reasoning models could be trained at a fraction of the cost previously assumed necessary. Zhipu’s GLM-5, released in February 2026, achieved the highest ranking ever by a Chinese laboratory on Artificial Analysis’s global open-source model benchmarks.
The cost-competitiveness of these models matters to investors for two reasons. First, it validates the commercial viability of domestic AI providers absent access to the most advanced semiconductor equipment subject to US export controls. Second, lower inference costs accelerate user adoption, expanding the addressable market and accelerating the penetration narrative that underpins current valuations.
3.4 Investor Sentiment Comparison
Dimension United States China
Primary investor concern Disruption of incumbent profit pools Pace of market penetration
Dominant sentiment Risk-averse; scare trade rotation Risk-on; growth and catalyst chasing
Listed LLM exposure Extremely limited (OpenAI, Anthropic unlisted) Growing (MiniMax, Zhipu, others)
Foreign competition risk High (global open market) Low (regulatory insulation)
Benchmark narrative Mixed; concerns about commoditisation Positive; GLM-5 global top ranking
Halo effect drivers NVIDIA infrastructure boom US private valuations, DeepSeek momentum
Key risk AI accelerates faster than anticipated Earnings fail to justify re-rating
- Singapore: The Strategic Nexus
4.1 Singapore’s Role as a Financial Intermediary
Singapore’s financial sector occupies a structurally distinctive position with respect to this divergence. As the region’s preeminent asset management and wealth hub, Singapore-based institutions are simultaneously allocating capital to both US technology companies and Chinese AI stocks. Firms such as Saxo Markets and Allspring Global Investments—whose analysts are among the most widely cited in commentary on the China-US AI divergence—are headquartered or regionally anchored in Singapore, reinforcing the city-state’s role as an analytical and capital intermediation hub.
4.2 Capital Flow Dynamics
Singapore functions as a conduit for regional and global capital accessing Chinese AI equities, particularly through Hong Kong-listed vehicles. The recent listings of MiniMax and Zhipu on the Hong Kong Stock Exchange are accessible to Singapore-based institutional and retail investors in ways that US-listed AI infrastructure stocks may not fully replicate. As Chinese AI listings multiply and the Hong Kong market deepens its AI sector, Singapore’s fund management industry is well-positioned to serve demand from both regional and international allocators.
The city-state’s neutral geopolitical positioning—maintaining active economic relationships with both Washington and Beijing—also allows Singapore-based firms to navigate the US-China technology bifurcation with greater flexibility than counterparts domiciled in either country. This is a material competitive advantage in an environment where technology investment is increasingly shaped by regulatory and geopolitical constraints.
4.3 Singapore’s Domestic AI Sector
Beyond its intermediation role, Singapore is building its own AI capacity. Government-backed initiatives including the National AI Strategy 2.0 and investments through agencies such as the Economic Development Board and the Infocomm Media Development Authority are positioning Singapore as a regional AI research and deployment hub. Singaporean enterprises—across financial services, logistics, healthcare, and public administration—are early adopters of both US and Chinese AI tooling.
The divergence in investor sentiment has direct implications for Singaporean companies. Firms deploying AI to drive operational efficiencies may find that market participants assess them more through the Chinese ‘penetration’ lens than the US ‘disruption’ lens, at least in the near term. Conversely, Singapore-listed firms in sectors considered vulnerable to AI disruption—financial services, professional services—may face valuation headwinds as global scare trade dynamics diffuse into regional markets.
4.4 Risks and Vulnerabilities
Singapore’s exposure to this divergence is not without risk. First, as a small open economy, Singapore is highly sensitive to shifts in global capital allocation patterns. A synchronised re-rating of Chinese AI stocks—should earnings growth fail to match investor optimism—could transmit negative sentiment rapidly through Singapore-based funds with Chinese AI exposure.
Second, Singapore’s financial sector faces its own version of the disruption narrative. Wealth management, legal services, and financial advisory—sectors in which Singapore has significant employment and export income—are precisely the industries most exposed to AI-driven cost compression in the US context. Whether Singapore’s market prices this risk as aggressively as US counterparts do remains an open empirical question.
Third, Singapore’s geopolitical neutrality, while currently an advantage, requires active management. The US export control regime on advanced semiconductors creates compliance obligations for Singapore-based companies and financial institutions with exposure to Chinese AI supply chains, requiring careful navigation of an increasingly complex regulatory landscape. - Analytical Framework: Why the Divergence Persists
5.1 Structural vs. Cyclical Drivers
The China-US investor sentiment divergence is better characterised as structurally driven than cyclically driven. Cyclical explanations—such as differences in interest rate environments or short-term earnings momentum—do not adequately account for the depth or consistency of the divergence. The structural explanations are more compelling: market insulation, different stages of AI penetration, distinct competitive dynamics, and divergent availability of listed investment vehicles all contribute to a fundamentally different investment calculus.
This distinction matters for investors assessing the durability of the divergence. If structural, the divergence may persist for several years until Chinese AI companies generate sufficient earnings to test valuations, or until US incumbent disruption becomes visible enough in reported financials to confirm or contradict the scare trade hypothesis. If cyclical, a normalisation could occur more rapidly.
5.2 The Earnings Test
The most significant near-term risk to the Chinese AI re-rating is the earnings test. Current valuations for MiniMax and Zhipu are priced for significant future growth; they are not, for the most part, justified by current revenue or profitability. The Jefferies analyst team noted in their February 13 note that ‘there is upside to China AI valuations’ while simultaneously cautioning that the re-rating may prove difficult to sustain if earnings growth fails to keep pace with investor optimism.
This mirrors the challenge facing US AI infrastructure stocks, where NVIDIA’s reported earnings have thus far validated elevated valuations, while software incumbents have yet to report sufficient AI-driven revenue compression to confirm the scare trade thesis. Both markets are, in different ways, awaiting the realisation of anticipated outcomes.
5.3 DeepSeek as a Systemic Variable
DeepSeek occupies a unique position in the global AI investment narrative. Its R1 model, released in early 2025, simultaneously undermined the assumption that frontier AI required prohibitive capital expenditure and validated the thesis of Chinese AI competitiveness. Its anticipated next-generation model release in early-to-mid 2026 is widely expected to serve as a further catalyst for the Chinese AI sector, with potential read-through effects on global AI infrastructure spending assumptions.
For Singapore investors, DeepSeek represents both an opportunity and a complication. As a private company, direct investment exposure is unavailable; the benefit accrues instead to listed comparables and downstream enterprise adopters. The cost-efficiency narrative also supports the broader AI penetration trade in markets beyond China, including Southeast Asian markets where Singapore-based firms often serve as capital conduits. - Conclusions and Strategic Implications
6.1 Summary of Key Findings
This case study has established that the divergence in AI investor sentiment between China and the United States is structurally grounded rather than ephemeral. The principal structural drivers are: the regulatory insulation of China’s domestic AI market from foreign competition; the scarcity of publicly listed LLM developers globally, which confers a premium on Hong Kong-listed Chinese AI stocks; the demonstrated cost-efficiency of Chinese AI models; and the differing stages of AI penetration across the two markets.
Singapore is positioned at the intersection of these dynamics as both a capital intermediation hub and a domestic economy navigating the same AI transformation pressures. Its financial institutions are analytically influential, its capital markets are well-connected to both Hong Kong and US equities, and its geopolitical neutrality provides operational flexibility unavailable to counterparts in either the United States or China.
6.2 Implications for Different Stakeholders
For institutional investors based in Singapore, the divergence creates opportunities to construct differentiated portfolios that capture AI growth through Chinese pure plays while managing disruption risk in US-exposed positions. The intermediation role of Singapore-based asset managers is likely to grow as regional and global allocators seek expertise in navigating the bifurcated AI investment landscape.
For policymakers, the divergence highlights the importance of Singapore’s AI strategy in both its domestic and international dimensions. Ensuring that Singapore-based enterprises are effective adopters of AI—and that the financial sector in particular manages disruption risk proactively—will be critical to maintaining Singapore’s competitive position as the AI transition accelerates.
For researchers, the China-US AI investment divergence offers a rich empirical case for examining how market structure, regulatory environment, and geopolitical context shape the pricing of transformative technology. The earnings test over the next 12 to 24 months will provide important evidence about whether structurally distinct markets ultimately converge on similar valuation frameworks as AI capabilities and adoption mature.
6.3 Outstanding Questions
Several questions remain open and will be important to monitor. Whether Chinese AI model developers can translate benchmark competitiveness into durable commercial revenue remains unproven. The extent to which Singapore’s financial sector faces the same disruption pressures as its US counterparts—and when that risk becomes visible in valuations—is uncertain. Finally, the geopolitical trajectory of US export controls and their effects on Chinese AI development timelines continues to be a source of significant uncertainty for all market participants.
References and Sources
Bloomberg / The Straits Times (February 22, 2026). ‘China defies global AI scare trade as investors chase winners.’ Published by SPH Media Limited.
Jefferies Financial Group, Edison Lee et al. (February 13, 2026). Research note on China AI sector valuations.
Artificial Analysis. (February 2026). Global LLM Benchmark Rankings.
Saxo Markets, Charu Chanana. (February 2026). Commentary on China-US AI investment divergence.
Allspring Global Investments, Gary Tan. (February 2026). Commentary on China AI competitive landscape.
— End of Case Study —