In the latest episode of The Investor’s Guide to Asia, portfolio managers Joseph Zhang and Terence Tsai argue that the AI investment story across mainland China, Taiwan, and South Korea - and elsewhere - has further to go, even if some elements look more stretched than a year ago.
Asia sits at the heart of the global AI supply chain, and its stock markets have been reaping the benefits. Yet a correction in July raised questions about whether the trade had become overcrowded and whether the region’s leading beneficiaries would continue to deliver. This month’s podcast brought together two of our portfolio managers who are closely following the theme.
Are we in an AI bubble yet?
A chips industry expert, technology portfolio manager Terence Tsai has invested in the AI boom for some time and each year he runs through the same process: a series of research trips aimed at finding the flaws in his investment thesis and a reason to step back. Yet time spent in Silicon Valley and Taiwan in the past three months has only left him more convinced of the case for the companies providing the infrastructure for the boom.
“We were already positioned heavily in the enablers of the boom,” he says. “But every time I finish my trips, I end up buying more infrastructure plays.
“It's quite unusual to have overwhelming positive feedback on the state of demand. Usually there are ebbs and flows, and usually not everyone is as positive about a trend or a theme. But when I speak with companies, when I speak with start-ups, with hardware companies, with labs, everyone is short on compute.”
There are signs that parts of the market are “getting stretched”, Tsai says, with more companies relying on debt and equity financing rather than just operating cash flows to fund capital expenditure. But the technology’s adoption and the earnings growth across the sector have continued to exceed expectations.
“AI has clear intrinsic value - it’s influencing many aspects of daily life while companies participating in the infrastructure build-out are producing genuine profits and cash flows,” multi-asset portfolio manager Joseph Zhang also tells the podcast.
AI-related stocks have delivered strong performance this year, Zhang argues, but he points to some detailed comparisons of valuations that suggest growth in earnings has left tech companies still very reasonably priced.
At time of recording, the MSCI All Country World IT Index traded on a 12-month forward price-to-earnings ratio of 17.4x, compared with 16.4x for the broader MSCI All Country World Index. The ratio of the IT sector's valuation to that of the broader market is near a multi-decade low, indicating that the sector’s relative valuation premium is modest by historical standards.
“This is an earnings-driven upcycle,” says Tsai. “You're not buying these stocks with the promise of something fuzzy in the future. These earnings are coming now, large and fast.”
South Korea and Taiwan: Different routes into the AI trade
Tsai says traditional signs of a semiconductor downturn, such as rising inventories, shorter lead times, and declining factory utilisation, have yet to emerge. Instead, inventories remain low, lead times are lengthening, and fabrication plants are operating at full capacity. In Asia, many of the principal beneficiaries are found in Taiwan and South Korea.
For an asset allocator like Zhang, the two markets offer different forms of exposure to the AI theme. South Korea is relatively concentrated in memory-chip manufacturers, while Taiwan has a broader ecosystem encompassing everything from advanced chipmaking to server components, frames and cooling systems. Taiwan therefore provides more diversified exposure to the AI infrastructure supply chain, albeit at higher valuations.
July’s correction in those markets, says Zhang, was driven primarily by technical factors rather than deteriorating fundamentals. He attributes it to crowded positioning, regulatory scrutiny of leveraged exchange-traded funds, and stress among some market participants. Company fundamentals have nevertheless remained broadly intact, and the contraction in valuation multiples created more attractive entry points for long-term investors.
China's AI catch-up story
Both managers are paying increasing attention to China's efforts to develop alternative supply chains as the country seeks to compete with the US in chip manufacturing and frontier AI models. Zhang views Chinese technology holdings as a source of diversification and a hedge should China narrow the gap with the US more quickly than expected.
“I always think the main risk to a US-led AI narrative would be if China catches up,” says Zhang. “What if China can one day produce leading-edge chips and memory products? People would quickly rethink the supply-demand dynamic. It is a risk that I watch very closely.
“Right now, China is catching up, but it’s not there yet.”
Tsai says US export controls and restrictions have forced Chinese companies to develop alternatives and there is growing evidence that domestic substitution is working, both in AI models such as DeepSeek and Kimi, and semiconductor hardware and manufacturing. Chinese firms are not yet leaders in the most advanced technologies, but they are becoming increasingly competitive across much of the supply chain. This is supporting the emergence of domestic champions that could deliver long-term growth and offer lower-cost solutions to global customers.
“People used to have this misconception that China was about making cheap, low-end and commoditised products,” says Tsai. “But increasingly, [Chinese companies] have shown that in the face of constraints and challenges, they're able to innovate around that.”
What could derail the AI story?
While rising bond yields and higher borrowing costs may create headwinds for broader markets, Zhang believes the impact on AI-related investment is likely to be limited. Hyperscalers continue to treat AI as a strategic priority, while many governments across Asia are directing fiscal support towards AI.
Another potential challenge is growing resistance to data centre construction, driven by concerns over electricity consumption, water usage, and land availability. Tsai sees this as a genuine execution risk to the AI build-out. However, he argues that demand remains strong and that the industry's primary challenge is expanding supply quickly enough to keep pace.
Looking further ahead, while acknowledging concerns about AI's societal impact, they both argue that history suggests transformative technologies ultimately create more opportunities than they destroy.
