US tightens AI chip curbs: Why US and China tech stocks face new risks
Key takeaways:
- China AI stocks are diverging: Domestic chipmakers such as SMIC and Hua Hong are benefiting from the self-sufficiency narrative, while AI developers face higher computing costs.
- US and Asian tech face indirect risks: Nvidia's immediate China exposure may be contained, but restrictions could affect future demand for AI chips, cloud infrastructure and memory.
- Regulation is becoming a bigger AI investment risk: Washington is extending restrictions from physical chip sales to remote computing access, with concerns that Chinese AI models could eventually face restrictions too.
What's changed?
Washington has introduced new restrictions targeting Chinese companies' access to advanced US AI chips through overseas data centres, including facilities in Southeast Asia.
The rules introduce additional licensing requirements affecting data centres and advanced chip shipments to Singapore, Malaysia, Thailand, Indonesia and Brazil, with exemptions for qualifying US companies.
This is a meaningful escalation because restrictions are moving beyond who can buy advanced chips to who can access their computing power remotely.
For overseas data-centre operators reliant on Chinese customers, this could mean weaker utilisation, pricing pressure and delayed expansion. Chinese AI developers, meanwhile, could face rising costs and slower development timelines.
What does this mean for China AI stocks?
The market reaction highlights an important distinction: what benefits China's domestic chipmakers may simultaneously hurt its AI developers.
Domestic semiconductor manufacturers such as SMIC and Hua Hong have attracted buying interest as investors anticipate greater investment in local manufacturing and self-sufficiency.
However, the impact across Chinese technology is likely to vary:
- Domestic chipmakers: Could benefit from stronger demand and policy support, although manufacturing constraints and expensive valuations remain risks.
- AI developers: Companies such as Z.AI and MiniMax could face higher model-training costs and reduced access to advanced computing.
- Internet platforms: Alibaba, Tencent and Baidu have diversified businesses and greater financial resources, but rising computing costs could still weigh on AI returns.
The result is likely to be increasing divergence within China technology rather than a uniform sector-wide rally or selloff.
Explore our China Tech & AI Shortlist for stocks and ETFs across China's semiconductor, AI and automation value chains.
What about Nvidia and other US tech companies?
For Nvidia, the bigger risk may be future demand rather than an immediate earnings hit.
China and Hong Kong accounted for approximately US$7.9 billion, or 8.2% of Nvidia's revenue, in its second quarter of fiscal 2027. However, this may not capture the full economic exposure to Chinese demand through overseas intermediaries and data centres.
Historically, Singapore accounted for as much as 22% of Nvidia's quarterly revenue under its previous billing-location reporting methodology, although much of that revenue was associated with products shipped elsewhere. Nvidia has since shifted to reporting revenue based on customer headquarters, making comparisons more difficult.
While Nvidia's current guidance already assumes zero Data Center compute revenue from China, the latest restrictions raise questions about how much future demand from overseas data centres could be affected.
For other US technology companies:
- AMD and Broadcom: Could face weaker demand for AI accelerators, networking equipment and related infrastructure.
- Microsoft, Amazon and Alphabet: May face additional compliance requirements, although diversified customer bases and exemptions offer some protection.
- AI infrastructure providers: Operators dependent on Chinese customers could face lower utilisation, weaker pricing and financing challenges.
The immediate revenue exposure is difficult to quantify, but the broader concern is that tighter restrictions could reduce the future addressable market for AI infrastructure.
Could this be the beginning of a broader regulatory push?
Potentially, and this may be the more important longer-term implication for investors.
The restrictions suggest that US technology policy is expanding beyond physical chip exports towards controlling access to AI computing power.
The next potential frontier could be AI models themselves.
According to Nikkei Asia, Chinese developers are increasingly concerned that Washington could restrict Chinese open-source and open-weight AI models on national security grounds.
This matters because lower-cost Chinese models have gained international users, including American businesses.
If such restrictions materialise, they could:
- Limit global adoption of Chinese AI models, reducing commercialisation opportunities for developers such as DeepSeek, Z.AI and MiniMax.
- Raise AI adoption costs for businesses globally, including American companies increasingly using cheaper Chinese models for coding, automation and routine tasks.
- Slow the next phase of AI's economic value creation. As the AI investment cycle shifts from building infrastructure towards adoption, affordable models are crucial for businesses to scale AI applications and generate measurable productivity gains.
- Accelerate fragmentation, with separate US and Chinese technology ecosystems potentially reducing competition, efficiency and innovation.
A broader crackdown is not inevitable. Washington must balance national security concerns against the costs to American technology companies and innovation. Nevertheless, regulation is becoming another variable investors need to consider alongside earnings, valuations and AI spending returns.
What should investors watch next?
Three factors will be particularly important:
- US-China negotiations: Whether the upcoming Trump-Xi meeting offers any clarity on technology restrictions.
- China's domestic AI progress: Whether local chipmakers can narrow technology gaps and whether developers can maintain competitiveness despite higher costs.
- Global AI infrastructure demand: Whether restrictions merely shift investment across regions or begin to reduce overall spending expectations.
Bottom line: AI is becoming a more selective investment opportunity
China's AI ambitions are not disappearing, but the cost of achieving them is rising.
Domestic semiconductor suppliers could benefit from the self-sufficiency push, while AI developers face greater uncertainty around costs and competitiveness.
For global investors, the message extends beyond China. AI remains a structural growth opportunity, but earnings quality, access to technology, regulatory exposure and returns on investment will increasingly determine the winners.
The AI trade may not be ending, but the assumption that every part of the value chain benefits equally is becoming harder to defend.