Deglobalisation Takes Centre Stage in Semiconductor Land

Globalisation and free trade are almost dead: TSMC founder Morris Chang
Listen to this story

In light of the semiconductor industry witnessing a giant geopolitical shift, TSMC founder Morris Chang, at a recent event in Phoenix, Arizona, famously said, “globalisation and free trade are almost dead”. Chang’s statement follows a movement towards the regionalisation of fabs. 

Speaking about the impact of nativisation on the global semiconductor industry, Naveed Sherwani, CEO at RapidSilicon, interpreted “globalisation is dead” to mean that chip prices will have a consistent upward trajectory. 

Until now, progress in technology has been a measure of upholding Moore’s law, where more and more performance on less and less price year after year has been the norm. However, this was only possible because of the socio-economic environment, which encouraged the global collaboration of companies in materials, equipment, design/IP, manufacture and pack/test. But, with economic nativism, he adds that two different supply chains will have two different suppliers—both underutilised and more expensive to operate.  


Sign up for your weekly dose of what's up in emerging technology.

Towards regionalisation

Taiwanese chipmaker TSMC contributes majorly to its 53% market share in the global foundry market. But, the company, of late, has been accelerating its plans to look beyond Taiwan, to the West. The 1,000-acre plant announced in Arizona will work in two phases and is expected to bring $10 billion in annual revenue for TSMC. In addition, the manufacturing giant announced the construction of its fab in Japan, initially housing 22/28 nm process technology capacity, as well as looks to be the first to set up a fabrication plant in Europe, with reports of officials eyeing sites in Dresden, Germany. 

Like TSMC, besides their long-term plan to invest in Korea, Samsung announced a $17 billion investment to build a fab in Texas as part of Samsung’s goal to increase foundry production capacity to up to three times by 2026. Earlier, Samsung also teased that it is considering building 11 semiconductor manufacturing fabs in Austin, Texas, at an estimated cost of $200 billion. 

Download our Mobile App

But, before their foray into multiple new locations, thereby diversifying their supply chain apparatus, companies were operating multiple fabs, called Gigafabs, often placed at a single location. These Gigafabs are seen across the board. 

For instance, Samsung Electronics, the second-largest global pure-play foundry revenue contributor, has wafer fab complexes in South Korea (the P3 plant alone produces about 100,000 12-inch sheets of wafer per month). Kioxia, in shared ownership with Western Digital, owns a collection of fabs in Yokkaichi, Japan. Similarly, Intel announced two new fabs in Arizona under the CHIPS Act, resulting in a trio of fabs in the desert of the Grand Canyon State. 

The multiple fab clusters allow companies to access a strong talent pool, a developed ecosystem of supply chain partners, and a solid infrastructure to meet the needs for utilities like electricity, water, and other services. Bigger fabs also lead to faster cycle times (the time it takes to run through a single layer on a wafer) since wafer batches are immediately and automatically routed to a different machine each time one goes down. 

Foundry giant TSMC has three “Gigafabs” located in Taiwan—namely, Fabs 12, 14 and 15—with a capacity of around 150,000 wafers per month each. In comparison, the Arizona facilities will be 20000 wafers per month, the minimum efficient scale. 

Thus, as an effect of regionalisation, we are seeing companies move to multiple subprime locations, leading to the creation of more small fabs, as opposed to large gigafabs. 

Why diversify? 

SemiAnalysis’ Dylan Patel discusses four aspects that show why the world is heavily dependent on East Asia. 

(i) The lithography tool, considered to constitute a major chunk of a fab investment, is said to be a monopoly of the Dutch multinational corporation ASML. However, what needs to be added to the conversation is that these tools are only useful with the supporting Japanese tools and chemicals. 

(ii) Japan, South Korea, Taiwan, China, and Singapore generate more than 90% of the world’s DRAM and NAND. In addition, Taiwan and South Korea are the only countries that produce all chips on nodes with a theoretical density of more than 100 million transistors per square millimetre.

(iii) The top 10 foundries have their largest manufacturing facilities in Asia, specifically in Taiwan, China, South Korea, and Singapore. 

(iv) More than 90% of Outsourced Assembly and Testing (OSAT) are done in East Asia, while more than 60% of chip packaging is in China. 

But, on the flip side, Taiwan suffers from geopolitical and seismic difficulties. On the brink of a Chinese invasion, it has found little support from the West. As a result, investors are increasingly sensing the ‘Taiwan risk’. Moreover, the US ban on certain exports of chip technology to China will cause global collateral damage to the entire global technology ecosystem, which includes chip design, tooling, and raw materials. 

In his book Chips War, academician Chris Miller notes that the globalised condition is facing what can be called “weaponised interdependence”, where interconnected interests of the economic giants produce “new arenas for competition” rather than resolving disputes and promoting cooperation. Therefore, furthering investment in local chip manufacturing is one way to offset these risks – something that many countries are trying to do. 

A case for self-reliance

Bharat Belavadi, senior director & India head, Advanced Analytics Office at Western Digital, told AIM that semiconductor manufacturing is a highly complex and capital intensive industry, and with the explosion of semiconductor usage, companies are under a lot of stress on the supply chain. He adds, “It takes nearly 5-6 months for a wafer to get manufacturing, undergoing nearly 1700 steps, with the wafer moving about 30 miles within the fabrication unit during production.”  

According to Belavadi, “Deglobalisation will help in recovering the huge influx of demand, and countries becoming self-reliant in producing critical components.”

Therefore, countries and its administration no longer want to depend on other countries and suppliers to fulfil their demand for semiconductors. Consequently, the semiconductor producers also find it easier to diversify their supply chain outside China and Taiwan because of highly-incentivised policies drafted by nations elsewhere. For example, the CHIPS Act in the US has incurred $200 billion in private investments in the domestic semiconductor sector. India, too, has joined the semiconductor bandwagon with the revised semiconductor policy that came into effect in September this year. Both state and central government incentives will greatly support companies with infrastructure and capital needs. 

More Great AIM Stories

Ayush Jain
Ayush is interested in knowing how technology shapes and defines our culture, and our understanding of the world. He believes in exploring reality at the intersections of technology and art, science, and politics.

AIM Upcoming Events

Regular Passes expire on 3rd Mar

Conference, in-person (Bangalore)
Rising 2023 | Women in Tech Conference
16-17th Mar, 2023

Early Bird Passes expire on 17th Feb

Conference, in-person (Bangalore)
Data Engineering Summit (DES) 2023
27-28th Apr, 2023

Conference, Virtual
Deep Learning DevCon 2023
27 May, 2023

3 Ways to Join our Community

Telegram group

Discover special offers, top stories, upcoming events, and more.

Discord Server

Stay Connected with a larger ecosystem of data science and ML Professionals

Subscribe to our Daily newsletter

Get our daily awesome stories & videos in your inbox

A beginner’s guide to image processing using NumPy

Since images can also be considered as made up of arrays, we can use NumPy for performing different image processing tasks as well from scratch. In this article, we will learn about the image processing tasks that can be performed only using NumPy.

RIP Google Stadia: What went wrong?

Google has “deprioritised” the Stadia game streaming platform and wants to offer its Stadia technology to select partners in a new service called “Google Stream”.