For AI adopters, diversification across US and alternative model sources is increasingly essential. For AI investors, investing in what the US and China each do best is also strongly advised. Read on for more information. 
 

Source: AI Index Report 2026, Stanford University.

The Strait Times recently featured an article, "Is Singapore ready for an AI outage?", highlighting a structural vulnerability related to AI. We’d like to share a few key takeaways from this piece. Hope you find them useful.

• Concentration risk. Singaporean businesses and government services increasingly rely on a handful of American AI models. This creates an “all eggs in one basket” risk where a sudden service cut could halt critical operations in the city-state – from banks to hospitals. 
   

• AI as a US national security concern. The US is clearly tightening foreign access to its AI models, including suspending foreign access to Anthropic's Fable 5/Mythos 5 and OpenAI's GPT 5.6. This signals that the White House now views frontier AI as a national security asset, which could lead to export control and eventually cutting off non-US access. 
   

• Diversification rather than isolation. The article suggests Singapore doesn’t need to build frontier models from scratch. Instead, resilience comes from owning the application layer, allowing systems to swap models seamlessly and to strengthen open-source and regional models as backups.

Our Take

• Weaponization of AI technology. The latest LLM restrictions suggest that frontier AI models are no longer just commercial tools, but national security assets. This both validates and extends our 2026 top prediction of the "US-China AI Arms Race," where AI becomes a core instrument of strategic competition between countries.
   

• "Compute sovereignty" bottleneck for rest of the world. The recent US limits on “bleeding edge” models (Fable 5, Mythos 5, GPT-5.6 and counting) further demonstrate how Washington plans to control AI advancement as a geopolitical chokepoint. For international economies, overreliance on US frontier models could become an unexpected liability rather than a competitive advantage.
   

• Bifurcated AI landscape demands diversification. The world is rapidly moving toward a fractured AI ecosystem, with US-controlled “walled garden” models subject to policy whims, and a parallel open ecosystem of Asian alternatives (mostly Chinese). As we noted before, budget-friendly, more efficient Chinese AI models are gaining global traction. Corporates should prioritize diversification across both ecosystems, in order to mitigate the risk of sudden, policy-driven blackouts and capture the best each side can offer.
   

• No need to take sides – just invest in the best of each. For investors, we also advise a diversified, "two-pronged" strategy. There is no need to take sides, but simply invest in what each of the US and China does best. For example, the US continues to dominate in AI computing hardware, while China is increasingly competitive in low-cost, high-efficiency models. Therefore, our Core Recommendations include Nvidia (NVDA) and Broadcom (AVGO) as leading US chip suppliers, as well as Tencent (0700.HK) and Alibaba (9988.HK / BABA) as Chinese model enablers.