Market Update: Positioning in AI ecosystem in Asia amidst volatility in semiconductors

Highlights: The pullback in semiconductor and memory stocks looks like rotation rather than capitulation, driven by investors reassessing whether profits growth could meet the elevated expectations. We still see the structural AI theme as intact and maintain positions across the AI ecosystem in Asia including power, infrastructure, and industrial automation. Model competition is intensifying and pricing pressure is rising as the subsidised AI era fades, pushing providers towards monetisation via offerings like Model-as-a-Service (MaaS). In China, we have seen renewed investor’s preference for biotech, internet platforms, hyperscalers and EVs.

• With the sharp increase in memory prices, the profits of memory makers have continued to rise rapidly. But some investors may question whether such growth will meet the elevated expectations. South Korea equities are going through period of remarkably high volatility, with domestic retail investors as net buyers while foreign investors being net sellers. Leverage indicators are off their peaks but still elevated. Regulators are tightening rules to reduce leverage and dampen volatility
• We see the current pullback in semiconductor and memory stocks as more of a rotation than capitulation. The strong AI capex, which is projected to increase from below USD400 billion in 2025 to surpass USD1 trillion in 2028e, should help businesses bring in new revenues. Asia is becoming the epicentre of the global data centre boom, with regional capacity expected to more than double by 2030 to around 40 per cent of global capacity. This build-out should boost demand across the data centre supply chain and power ecosystem, including chips, semi equipment, cooling, servers, commodities, onsite power generation, and energy storage. This is the core thesis of our High Conviction theme, Asia’s Data Centre Boom
• China’s token economy is emerging quickly, and the Total Addressable Market (TAM) for China’s LLMs could exceed USD150 billion by 2030, creating attractive upside for AI revenues – especially for AI hyperscalers that can monetise across established ecosystems. Chinese foundation models are gaining traction thanks to cost competitiveness and a narrower performance gap versus frontier models. Beyond AI, China’s scale advantage is showing up in advanced manufacturing – particularly EVs (with autonomous driving) and energy storage – and strong biotech capability, supporting a focus on leaders across the AI and advanced manufacturing value chains. We reiterate our High Conviction Theme, China’s Tech and Manufacturing Leaders