The AI trade: do geopolitics matter?

Plus, scepticism on the dealmaking rebound

AI and geopolitics AI investors have mostly brushed off geopolitical tensions as the trade runs higher on hopes of innovation and future productivity benefits. But the market tide is slowly shifting on this front, particularly with regard to US-China relations. Fridays sell-off is a key example: Nvidia and AMD fell nearly 5 per cent and 8 per cent, respectively. Other examples from this year include the DeepSeek-fuelled US tech sell-off in January, and when Nvidia shares fell in September after China banned domestic tech companies from buying its chips.

Theres an argument that takes this view one step further that predicting breakthroughs is a fools errand, and AI investors would be better off making judgments based on geopolitics and supply chain chokepoints. As Henry Wu from Alpine Macro explains: At the bleeding edge of innovation, breakthroughs are nearly impossible to predict...Value creation will be shaped by where bottlenecks occur and how a G-2 US-China rivalry reshapes global technology flows Wu says companies at the bottleneck points in the AI supply chain such as lithography, memory and logic fabrication will be the near-term winners. These include SK Hynix, Samsung, TSMC, Nvidia, ASML and AMD, whose stocks have already notched, at minimum, double-digit growth year to date; their advantages across the supply chain arent a secret to investors. Its in the long term where the winners and losers could significantly change from what we currently expect, per Wu:

Over time, firms from US-allied countries will be squeezed by Chinese tech advancements and US reshoring...Hedge US AI investments with Chinese counterparts to reduce exposure to geopolitical risk Wus point is compelling, but only to an extent. As of now, no single country has a complete AI supply chain, as he shows in the following chart:

As acrimonious as relations are between the G2 in the AI arms race, it seems unlikely that China or the US can create the entire AI supply chain, or that the entire supply chain can be defined between the US and its allies versus China. The binary view feels misplaced. After all, the US allies identified in the chart are also highly dependent upon business with China.

Wu is wise to point out that China is also investing heavily in AI research and development, for which investors could benefit through exposure. Although there is the question of how much of the AI technology created can, or will be implemented, and by whom. The broader problem of implementation, even without factoring in geopolitics, might be more important for the future of AI investing. Jim Reid at Deutsche Bank notes that the US is great at making AI, not using it. In a recent Microsoft survey the US ranked just 23rd in terms of actual AI adoption, according to Reid:

The US may be the centre of the AI revolution, with 32 of the top 50 AI companies based in California alone, but it is a comparative laggard when it comes to actual users...It turns out that applying AI at work is a lot more complicated than initially expected. Generative AI adds a whole new layer of governance issues on top of the challenges of normal technology integration, let alone the problem of how to recruit AI talent and ensure that employees make the most of unfamiliar tools.