Emerging market monitor

In the 12-months since Chat-GPT was launched, investors have thrown themselves headlong into artificial intelligence (AI) stocks. Or, more precisely, into US-listed AI stocks.

Because, while Nvidia, which makes the chips that train AI computers, has soared 250 per cent over that period and US tech giants Microsoft and Meta have also experienced dizzying rallies, non-American firms key to the development of machine learning have been left trailing. Among the overlooked is a large group of Asian companies crucial to the AI supply chain.  Increasingly, they represent a missed investment opportunity.

That’s not to say it is wrong to regard US technology companies as the main beneficiaries of the latest advances in AI. They undoubtedly have the expertise, experience and financial firepower to transform the new technology into a huge commercial success.

The issue is that investors cannot afford to be complacent about the valuations these firms now command. There is a lot of future growth baked into those stock prices after all. All of which means that anyone looking to capitalise on the unfolding AI revolution will soon have to look beyond the increasingly expensive US tech sector.

Alternatives to Nvidia, Meta and co. abound. Yet perhaps the most promising source of investment returns are AI-related companies based in emerging Asia. These are the 40 or so listed businesses operating out of countries such as Taiwan, South Korea, China that manufacture almost all of the world’s AI chips and many of the AI-enabled products that are essential to the growth of the technology; they are providing the picks and shovels in this AI goldrush.

The AI investment opportunities we have identified fall into three categories.

To begin with, there are the companies that form part of the graphic processing unit (GPU) supply chain. These are the firms facilitating – and benefiting from – the growth of Nvidia’s specialist AI chips. In fact, all of Nvidia’s AI processors are produced, packaged and embedded into servers by Taiwan-based companies. As it boosts the processing power and memory of  its AI chips, Nvidia will increasingly rely on, among other things, the sophisticated processor packaging technology – known as chip-on-wafer-on-substrate, or CoWos  - that has been honed by Taiwan’s largest semiconductor group, TSMC.

Another hunting ground for AI-oriented investors is to be found among the companies that manufacture cheaper AI chips for Nvidia’s US competitors including Microsoft and Amazon’s own chip unit. The high prices that Nvidia’s processors command – and the limited supply of such products – are opening up opportunities for rival, low cost producers.

Among the most promising alternatives to Nvidia’s GPUs are processors known as Application Specific Integrated Circuits (ASIC). While high-end computing still requires GPUs, ASIC-powered servers can be deployed in other areas of AI development such as model inferencing, the process through which a trained AI model infers results from the analysis of live data. ASICs offer cost savings of around 10-20 per cent compared with Nvidia's AI servers. Delivery times are shorter too. Companies that specialise in the production of ASIC chips include Taiwan’s Alchip and Wiwynn. The third group of investment opportunities are Asian manufacturers of AI-enabled hardware.

The launch of a new breed of AI-enhanced smartphones and personal computers is expected to usher in an personal tech replacement cycle that could disproportionately benefit Asia-based hardware firms. According to some analysts, the growth in unit sales of AI enabled phones could increase at a compounded annual growth rate of some 80 per cent through to 2027 (see Fig 2).

AI’s expansion is already providing a significant boost to the revenues of tech companies in emerging Asia. TSMC's AI-related revenue, for example, is expected to more than double to over 10 per cent of the group total in 2024 while its Taiwan-based competitor Wiwynn's AI ASIC revenue is projected to increase from less than 10 per cent of its total to more than 30 per cent in 2024.

Crucially for investors, these companies trade at far lower valuations than the US’s AI behemoths. By our calculations, the median price-earnings ratio for the companies that make up the EM AI supply chain is 19.2, offering a considerable discount to US semiconductor stocks, which trade at some 27 times 12-month forward earnings.

Strong fundamentals in EM AI supply chain

Moreover, their earnings and sales prospects are also solid: the median market forecast points to annual revenue and profit growth of 14 per cent and 26 per cent respectively.

The upshot to all this is that investors have several options through which to capitalise on the advance of AI. While the US technology sector is clearly one of them, the eyewatering rally enjoyed by stocks such as Nvidia suggest investors may soon need to look at cheaper alternatives.

Emerging Asian tech companies are among the most promising.