US losing the Semiconductor Game?
Before we attempt an assessment of the Semiconductor supply chain process and the $52 billion investment by Congress possibly on production facilities and a 25% tax benefit for semiconductor manufacturers let us look at a few numbers out of China and in the US right at the offset and the beneficiaries in the Semiconductor space: In 2019, total announced Chinese investments in fabs exceeded $215 billion; this has gone up significantly over the last couple of years given China’s interest to develop competencies and reduce reliance on US, South Korea and Taiwan. Industry estimates that government financed fabs in China could number 70 or more by 2023 compared with roughly two dozen now.
Some of the expansion plans announced in the US by Samsung, Intel and TSMC total roughly $50 billion; this is merely for expansion of three factories. Why we have stated these numbers at the outset is to indicate the fact that $52 billion investment by US government is merely a drop in a pond. Apart from the domestic manufacturers like Qualcomm, Intel, NVIDIA, Advanced Micro, NXP Semiconductors and Broadcom, other existing players in the global industry could benefit from the investments announced as this may also strengthen and diversify existing relations with players like ASML, Applied Materials Inc etc.
The deputy secretary of Commerce, Don Graves insisted in September, the US’s efforts to expand its own manufacturing capacity will not lead to increased competition for South Korean companies. US just won't have the capacity, even if it moves as quickly as it'd like on the Chips Act, he said, referring to the proposed legislation that seeks to provide $52 billion in funding for the U.S. semiconductor industry.
From a game theoretic perspective, what this plan has the potential (mere potential, nothing more than that) is the ability to spur consistent future investments in technology to avoid the disaster that steel industry in America faced. To face the crude fact, America is way behind in the sector and with the Chinese laying on the investment thick and fast, it might be too little, too late for them.
There is limited capacity for manufacturing chips below 7mm which are majorly imported from Taiwan and South Korea. This is presumably where the majority of the $50 billion investments will be. SIA estimates that a disruption in the production of logic chips at foundries in Taiwan could result in nearly $500 billion in lost revenues for electronic devices manufacturers that depend on this supply.
There is still no clear indication as to what the US government’s plans are to expand and set up new capacities. They must develop competencies all over again in a sector it once championed. The American way of just walking in and taking what you want, which worked with Oil in the middle east will not work here; Best short-term solution would be to establish supply chains with South Korea, which they announced in September.
Some of the risks that encompass most of the identified threats to semiconductor supply chains: fragile supply chains; use of obsolete and generations-old semiconductors and related challenges for continued profitability of companies in the supply chain; customer concentration & geopolitical factors and electronics production network effects.
The following policy recommendations are designed to address the current semiconductor shortage and the risks identified:
1. Promote investment and collaboration, in partnership with industry, to address the semiconductor shortage.
2. Strengthen the Domestic Semiconductor Manufacturing Ecosystem through incentives.
3. Fund the Creating Helpful Incentives for Production of Semiconductors (CHIPS) for America provisions in 2021.
TSMC announced plans last year to build an advanced chip foundry in Arizona, a $12 billion investment with completion scheduled in 2024. The plant will produce 5 nm chips with a capacity of 20,000 wafers per month. Post the announcement TSMC is also contemplating expanding its existing facility in US to a capacity of around 120,000 wafers monthly, the first plant with capacity exceeding 100,000 outside Taiwan. TSMC faces no uncertain pressure to increase its US manufacturing, with the United States contributing around 60% of its revenues.
Samsung is considering a $17 billion investment to expand its production capacity in the United States and is expected to commence operations in 2023. Technical details on the plant expansion are unclear; one report indicates that the facility would be capable of producing at the 3 nm node. Intel announced in March 2021 that it will invest $20 billion to expand its manufacturing capacity through construction of two new fabs at its Chandler, Arizona campus. The fabs will not only serve Intel’s requirements but will also provide foundry capacity for fabless customers.
Increasing domestic semiconductor production capacity would help address supply chain vulnerabilities in all segments of the semiconductor supply chain. Relations with China not being at their best, and China’s market share expected to go up from 16% to 28% this decade, coupled with national security and possibly reducing technological edge, US needs to invest in more advanced chip manufacturing.