Taiwan Semiconductor Manufacturing Company (TSMC) and Samsung Electronics are slowly waking up to the actual long-term costs of backing a US government mission to jump-start its anaemic semiconductor manufacturing sector in the face of a perceived growing threat from China.
The US push for domestic chip manufacturing was spurred by rising trade friction with China driven by national security concerns and supply chain disruptions at the peak of the Covid-19 (coronavirus) pandemic, with many companies looking to reduce their reliance on imports from Asia.
TSMC CEO C.C. Wei acknowledged last week the initial costs of operating overseas fabrication sites are higher than in its home market due to a number of factors, pointing to the early stage of the international semiconductor sector compared with a mature market in Taiwan.
On its Q3 earning call, the CEO stated other key reasons are the smaller scale of production at new facilities and higher costs throughout the supply chain.
Wei explained “strategic pricing” and economies of scale driving costs down mean TSMC will be able to “manage and minimise the cost gap to maximise its return our shareholders”. He also noted it is working closely with government to secure support (aka subsidies).
Negotiations with unions in the US slowed construction of a chip plant in the state of Arizona, with production now scheduled to start in H1 2025 instead of 2024.
Media reports suggested TSMC faced challenges finding workers with the specific skills required: it trained many of the 1,100 local staff at facilities in Taiwan.
Volume chip production in Japan using 12nm and 16nm process technologies is scheduled to begin in late 2024. It hired 800 local employees, with the majority sent to Taiwan for hands-on experience.
Construction of a specialty site in Germany will begin in H2 2024 and production is targeted in late 2027.
Meanwhile, Samsung has seen the cost of a second chip plant in the US state of Texas jump by $8 billion to more than $25 billion, due mostly to higher-than-expected construction fees.
Sources told Reuters earlier in the year costs could rise even more.
The companies applied for billions in subsidies through the US CHIPS Act, which is designed to incentivise manufacturing to revitalise chip production in the nation after it’s share of global chip fabrication capacity declined from about 40 per cent in 1990 to around 12 per cent in 2020.
Authorities in Europe are also investing in chip production, aiming to double its share of global output by 2023.
Founder of industry blog Radio Free Mobile Richard Windsor stated earlier in the year chipmakers’ capex is being driven by geopolitics rather than economics, raising the question of whether the new manufacturing sites are actually needed within the timeframe they are being built.
He also cited data from Boston Consulting Group estimating the total cost of ownership for advanced logic chips in South Korea and Taiwan is 20 per cent lower than in the US, and 28 per cent to 37 per cent lower in China, depending on the willingness to transfer technology.
Following additional restrictions on the sale of AI chips to China last week, the US Semiconductor Industry Association explained while it recognises the need to protect national security, “broad, unilateral controls risk harming the US semiconductor ecosystem without advancing national security”, a comment echoing many recent statements
It is becoming increasingly clear shifting production out of Asia despite significant subsidies will likely raise the cost of making chips, which will be passed on to consumers.
Windsor has long argued such moves will slow long-term growth for the entire technology sector.