TSMC factory in Taichung’s Central Taiwan Science Park By Briáxis F. Mendes (孟必思) - Own work, CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=91599168
The US-China technology confrontation, likely to become more acute during this term of the new Biden Administration, centres on a perceived need on both sides to dominate the technologies of the future, such as artificial intelligence and quantum computing.
But the real battle of the present is over one of the most important technologies of today: semiconductors, and specifically the ability to manufacture them at advanced levels. Semiconductor manufacturing equipment (SME) and the ability to build advanced fabrication facilities at scale will be two of the most important determinants of national power going forward. This is due to the complexity of pulling together the requisite capital, talent, and market conditions needed to ensure sustained utilisation, growth, and profitability over an extended period to justify the large-scale investments needed.
Governments in the US and European Union have also woken to the realisation that the manufacturing of key semiconductors is now dominated by companies in Asia, and specifically, Taiwan, a growing geopolitical hotspot. Hence both sides of the Atlantic are rushing to try to incentivise Taiwan’s leading semiconductor manufacturing company, Taiwan Semiconductor Manufacturing Corporation (TSMC) and the producer of nearly 90 percent of semiconductors globally at advance levels, to bring its expertise to their shores. The motivation here is complex, because historically the market has determined where semiconductors are made, though governments have played a secondary role in this process by subsidising manufacturing via grants, tax credits, and access to land, water, and power at concessionary prices. The key inflection point that has been reached is that a complex global supply chain that had been optimised for economic efficiency over decades is now being restructured around a political logic that increasingly conflicts with the efficiency incentive.
The trifecta of company business model, technological and engineering capacity, and government subsidy has vaulted TSMC into its current global leading position. TSMC has demonstrated over the past decade its ability to master the most complex manufacturing processes at scale and drive continuous innovation in the manufacturing of advanced semiconductors. TSMC pioneered the foundry or fab model, focusing on the manufacturing process and turning advanced chip designs from ‘fabless’ semiconductor design companies such as Qualcomm, Nvidia, and Apple into finished products for incorporation into a variety of devices, from smartphones to data centres. TSMC’s secret sauce also includes building a massive industry alliance with trusted suppliers and customers, a merit driven ecosystem which drives broader innovation in the industry.
Meanwhile, China is tripling down on attempts to boost the capabilities of its domestic manufacturing sector, along with its semiconductor manufacturing equipment (SME) sector. The massive 2014 National IC Investment Fund has been supplemented several times and provides subsidies to dozens of fabs under construction in China. The Fund also holds stakes in China’s leading domestic foundry, Semiconductor Manufacturing International Corporation (SMIC). SMIC’s future ability to compete with leaders TSMC, Samsung, and Intel has been for the moment crippled by US restrictions on the sale of advanced tools to the company. Even with access to the most advanced tools, SMIC would still struggle to master some elements of manufacturing at the high end, even with the assistance of key former TSMC officials in its management chain. China’s SME sector is developing fast, but Chinese firms in this domain remain at least a decade or more behind their Western counterparts in most areas of tool development.
The question now facing governments in the US and EU is multifactorial and requires addressing four challenges: 1) how to handle massive Chinese government subsidies to the semiconductor industry, 2) how to handle restricting semiconductor use for military applications without disrupting supply chains, 3) the national security issue generated by the perceived concentration of manufacturing capacity on Taiwan, and 4) the enormous costs of long-term government subsidies for attracting industry leaders to build even a small size fab and providing all the inputs it requires in personnel, materials, and technology, over a sustained period. Overlaying the problem is the brittle and vulnerable nature of the global supply chains feeding and sustaining the industry, as highlighted by the major downstream and upstream shortages that have rocked the industry and its key consumers in the auto, medical device, electronics, and data centre sectors.
The first two challenges involve economic competitiveness and national security, while the second involve changing the calculus of actors driven by market forces and successful business models to concentrate production in a particular geographic area. Long-term solutions for both issues are interrelated but require new thinking among policymakers about ways to reduce and manage the competitive and national security components, which will in turn improve the chances that government intervention can effectively result in some rebalancing of the existing geographical distribution of manufacturing and supply chains.
The challenge for the US and EU governments, is defining the desired end state of policies that include both tightening national security controls around semiconductor and SME supply chains and reducing the risks of high levels of concentration of advanced manufacturing in Asia. Policymakers in both the US and Europe currently have no plausible scenario on where they would like to be with China in 2030 overall, including in the technology domain. So far, neither the Biden administration nor the European Commission has articulated what the strategy intends to achieve, other than broad concerns about China’s military modernisation. The March report by the US National Security Commission on AI (NSCAI), Chapter 14 of which is a detailed blueprint for increasing export controls on SME for China, recommends that the US government clearly state that principles that will guide policies to protect critical technologies, while providing greater clarity to industry. The somewhat stunning admission of the recommendation is that “no such framework currently exists.” The challenge is defining a national policy here for what is a sprawling global industry that includes China.
Ramping up support to address concentration of advanced semiconductor manufacturing in Asia
The US Congress will likely finally pass legislation this summer funding the CHIPS Act, which will provide over $30 billion for incentives to attract big players like TSMC and Samsung to onshore manufacturing in the US. TSMC has already committed to building an advanced fab in Arizona, but it remains unclear whether the firm will expand the facility to a so-called gigafab of the type it typically builds on Taiwan, allowing it to expand capacity to meet demand and scale effectively over its customer base. Over time, it is likely that the TSMC Arizona fab will become part of the US government’s “trusted foundry” program, allowing it to supply semiconductors for the US military.
The EU is also in the game, with internal markets minister Thierry Breton eager to lure a major such as TSMC or Intel to build or expand capacity in Europe. It seems unlikely that TSMC will agree to build gigafabs in both the US and EU—senior company officials, from CEO CC Wei to founder Morris Chang have spoken about their reluctance to site fabs outside Taiwan, where they enjoy major advantages—Taipei has also pursued a policy encouraging TSMC and other Taiwan semiconductor firms to keep their cutting-edge facilities on the island to reduce the chance of IP theft. Recently the Taiwan government also clamped down on recruiting activity from mainland China, claiming Chinese semiconductor companies were ramping up longstanding efforts to poach talent and steal trade secrets.
The problem for both Brussels and Washington is agreeing on some alignment on a complex long-term strategy that can address all of the issues at hand—such an approach needs to succeed in both effectively addressing the competitive challenge from China, while reducing the global economy’s reliance on Taiwan and TSMC—amid the volatile cocktail of growing US-China tensions, frequent typhoons and earthquakes, and the actual and potential economic impact of the shortages of the type seen recently with auto-related semiconductors that spread to other sector and has focused minds across the Atlantic on the issue. The worst outcome of the geopolitical forces now arraying around the US-EU-China-Taiwan nexus would be the precipitation of a military conflict over Taiwan, which would entail at the least a massive disruption of the semiconductor industry and global economy based on it.
By Paul Triolo, Eurasia Group, Paulson Institute
Article category: Digital Technology and R&D