Understanding Taiwan's Standing in The Global AI Supply Chain Today

Given the rise of global economic uncertainty following the Trump administration’s unilateral reciprocal tariff imposition and its looming introduction of the semiconductor tariff on its trading partners, the latest forecast of Taiwan’s GDP growth has certainly defied the gloomy outlook of the world economy. - AI image, for illustration purpose only

A piece of news that has gone unnoticed by the Malaysian media: Taiwan's gross domestic product (GDP) is predicted to grow at a whopping 7.37 per cent for the year 2025 ─ a level unseen since the 10.25 per cent growth 15 years ago. Adding to the disbelief is that this forecasted rate is much higher than the 4.45 per cent prediction made just four months earlier. Given the rise of global economic uncertainty following the Trump administration's unilateral reciprocal tariff imposition and its looming introduction of the semiconductor tariff on its trading partners, the latest forecast of Taiwan's GDP growth has certainly defied the gloomy outlook of the world economy.

At the same time, Wiwynn Chairwoman, Emily Hong, has, for the first time, officially confirmed that her company is the manufacturing and infrastructure partner for the YTL Group's AI data centre in Johor. This, in effect, puts to rest the market rumour regarding the Taiwanese company's involvement in the project, considering that it is the supplier to American hyperscalers such as Amazon Web Services, Microsoft, and Meta. Likewise, such news is hardly reported by the Malaysian media outlets, which focus on YTL's cooperation with Nvidia ─ the artificial intelligence (AI) giant, which also outsources its infrastructural manufacturing to its list of Taiwanese suppliers.

From these two pieces of news, it is clear that Taiwan is 'quietly' riding on the global AI wave as the manufacturing partner for the US, as well as companies cooperating with the American tech multinationals, with the likes of YTL. Yet, such a reality continues to be absent in public discussions about Malaysia tapping the global AI wave as a driver instead of just being a host of data centres. More importantly, there have been persistent calls for Malaysia to achieve a resilient supply chain for risk reduction, but no consideration is paid to Taiwan's indispensable position in the entire AI supply chain. By all means, such a discursive vacuum is certainly baffling for any observer of geoeconomics and supply chain security today. This begs a question: How much do we understand Taiwan's current standing in the global AI supply chain today?

Three Hard Facts

To begin with, there are three hard facts that are to be recognised. First, Taiwan's indispensability in the global AI supply chain spans beyond its role as the chip producer as in the case of Taiwan Semiconductor Manufacturing Company (TSMC). Rather, Taiwan has institutionalised itself as the supplier (manufacturer) of other AI infrastructures used in data centres: the servers used in data centres; the motherboards and power modules inside each server; the liquid cooling systems needed to operate the high-computing-power of AI data centres; and the server racks that integrate all servers into one system.

As stipulated within Morgan Stanley's AI supply chain report, released in March 2024, it is estimated that 78 per cent of Nvidia's AI servers in 2024 will come from seven Taiwanese companies, namely Quanta, Inventec/ZT Systems, Hon Hai, Wistron, Wiwynn, Giga-Byte and Asustek. From AI server design, rack production-integration to the provision of manufacturing services, these are the 'Big Seven' that drive Taiwan's electrical and electronic (E&E) exports to the US and countries hosting the American tech multinationals.

Second, Taiwan has a complete tech manufacturing ecosystem that is hard to replicate by any economy in the world, with the exception of South Korea. Be it IC design, chip manufacturing, IC packaging, research and development (R&D), E&E training or the wider AI infrastructural production, its economy has its presence in each of these domains. Such a feat is no coincidence but the outcome of 50 years of continuous investments, operations and innovations by the Taiwanese private and public players in higher education, regulatory and business sectors.

As revealed by Nvidia's founder, president and CEO Jensen Huang, back in 2024, there are altogether 59 Taiwanese entities (companies and universities) involved in the multinational giant's AI supply chain, to which it also controls 92 per cent of the world's graphic processing unit (GPU) market based on the data by Jon Peddie Research (JPR). It therefore makes sense when Huang clearly stated that, "Taiwan is where everything Nvidia does begins, (before) our partners and ourselves take it to the world." Apart from South Korea, which also plays an indispensable role in the supply of high-bandwidth memory (HBM) to Nvidia and with similar scale of tech manufacturing ecosystem as Taiwan's, it is arguably difficult, if not impossible, for other economies (including the US) to replace the latter in the entire AI supply chain.

Third, Taiwanese companies have also become major suppliers of the non-Nvidia AI supply chain today. In the latest gambles by Amazon and Google to dislodge Nvidia's monopoly of GPU market as well reduce their dependence on the AI giant's chips, both companies are in the midst of deploying the application-specific integrated circuit (ASIC) chips ─ which in turn, are installed within their respective central processing units (CPUs) and tensor processing units (TPUs) ─ to operate their own data centres as well as those of their customers'.

Once again, Taiwanese companies remain the manufacturing partners for both Amazon and Google. Whereas the TSMC is the supplier of the cutting-edge chips to both American clients, both Alchip Technologies and MediaTek have become the latest IC design companies from Taiwan, engaged by the two American tech giants. In all, these latest developments reflect another hard fact that Taiwanese companies continue to be the most reliable partners within the newly-emerged supply chains driven by Amazon and Google in their endeavours to carve out their own AI markets.

What's In It for Malaysia?

Be it the AI infrastructures led by Nvidia or those driven by Amazon and Google, all three hard facts have demonstrated that Taiwan's grasp of the AI supply chain is both extensive and deep. But unlike the Nvidia-led supply chain, which had been explored by Morgan Stanley, the complete landscape surrounding the varying roles of Taiwanese companies in the emerging ASIC supply chains driven by Amazon and Google is virtually unknown to the world.

To this, more academic and market research can be conducted to examine Taiwan's current standing in the global AI supply chain. This is particularly important to semiconductor economies like Malaysia, which is seeking to ride the AI wave but continues to be dependent on Taiwan for AI infrastructures that operate their data centres. Needless to say, such a dependency contradicts the goal of establishing a resilient supply chain as touted by the Malaysian government.

That said, instead of taking Taiwan as a competitor, to which it is not, a thorough examination of its E&E manufacturing experiences offers good lessons to Malaysia on how to build its presence in the current AI supply chain. It is time for us to study the case of Taiwan in the global AI supply chain today.

The writer is Senior Lecturer at the Institute of China Studies (ICS), Universiti Malaya and Advisor to Taiwan Chamber of Commerce in Malaysia (TWCHAM)

Source: www.nst.com.my