The twilight of America's chipmaking industry

The US semiconductor manufacturing industry has been in decline for many years. In the past decade, the advanced semiconductor manufacturing industry in the United States has moved overseas, and Intel is the only semiconductor manufacturing giant in the United States. With the development of semiconductor manufacturing technology, the research and development of advanced process nodes requires a huge amount of funds, which makes the advanced semiconductor manufacturing industry more and more centralized. That is, only a few companies can continue to maintain research and development, and other companies that cannot afford to invest a lot of money have to withdraw from the competition track of advanced semiconductor manufacturing and turn to other semiconductor manufacturing tracks with smaller investments.

In the landscape of global semiconductor manufacturing competition, we see that Asia is dominating the entire industry: among the companies that are currently actively investing in the most advanced semiconductor process nodes, there are two in Asia (Samsung and TSMC) and only one in the United States (Intel), and Intel's overall technology is lagging behind.

In the past five years, as the international situation has changed, the US government has also been trying to reverse its decline in the semiconductor field. Its main idea is to bring the semiconductor industry back to the United States through financial subsidies. In January 2021, the US Congress passed the CHIPS Act, which in principle provides $52 billion in subsidies to the semiconductor industry in the United States. However, the passage of the CHIPS Act does not mean that the subsidy funds can really be in place. The specific allocation of funds to subsidize the semiconductor industry still needs to be decided by this year's Congress. In other words, as long as Congress does not pass the CHIPS Act funding plan, the US subsidies to the semiconductor industry will not really begin.

The semiconductor industry was once optimistic about the financial subsidies of the CHIPS Act, and the $52 billion subsidy did attract many giants in the semiconductor manufacturing industry to increase their investment in the United States. Intel, a domestic semiconductor giant in the United States, was the first to respond, announcing that it would build a large-scale advanced semiconductor processing plant in Ohio, with an investment of $100 billion, targeting the most advanced 25A and 18A processes expected to be mass-produced in 2025. Samsung announced last year that it would invest $17 billion to build new semiconductor manufacturing facilities in Texas, and TSMC also began to build a plant in Arizona under the attraction of these financial subsidies. In addition to semiconductor fabs, GlobalWafers, one of the largest suppliers in the field of wafer manufacturing, also announced that it would invest $5 billion to open a wafer manufacturing plant in Texas. If it is officially completed, it will be the first wafer manufacturing plant in the United States in 20 years.

However, as the specific funding plan of the CHIPS Act has reached an impasse in the U.S. Congress, the ambitious plan for domestic semiconductor manufacturing in the United States may also become a bubble. The discussion of the funding plan of the bill in Congress has fallen into a partisan dispute. Although both Democrats and Republicans agree that a large amount of subsidies are needed for the domestic semiconductor industry in the United States, they still cannot reach a consensus on where the subsidy funds will come from. As Congress will enter a recess in August, reaching a consensus before August will be a key time node for whether the bill can be finally passed this year, and the time left for Congress is getting shorter and shorter.

Some of the aforementioned giants that were attracted by the bill and were preparing to invest heavily in U.S. semiconductors are also preparing another set of plans. Intel has indefinitely postponed the groundbreaking ceremony for its Ohio Fab, which was originally scheduled for July 22, and CEO Pat Gelsinger made it clear last week that if the CHIPS Act funding cannot be passed in time, Intel will cancel its large-scale expansion plan in the United States and will turn to building factories in Europe. GlobalWafers has also made it clear that if the USICA Act cannot be passed, it will cancel its factory construction plan in Texas. Although TSMC has not clearly stated the corresponding plan, the company has previously stated that the cost of building a factory in Arizona has exceeded expectations, and it is estimated that if the funding cannot be passed, the factory construction plan in Arizona will be reduced.

Intel CEO Pat Gelsinger tweeted last week urging Congress to pass the CHIPS Act funding plan in a timely manner, otherwise the Ohio factory would not be able to officially start construction.

As the semiconductor industry cycle moves from the shortage and overheating cycle that began two years ago to the demand decline cycle in the next few years, it is expected that more and more semiconductor manufacturers will consider cost and policy subsidies as important considerations for investment direction. If the United States, which is no longer cost-competitive, cannot pass an attractive subsidy plan, then overseas semiconductor companies will basically not consider building factories in the United States, and even domestic semiconductor giants in the United States will move further overseas. Intel CEO Pat Gelsinger's announcement that he is considering building a factory in Europe is a very clear signal that if the CHIPS Act funding is not passed, the US semiconductor manufacturing industry will further decline.

The US semiconductor manufacturing industry was once glorious. When the semiconductor industry was just emerging (in the 1950s), almost all semiconductor manufacturing was completed in the United States, and Intel also led the rapid iteration and development of global semiconductor manufacturing technology at the speed of Moore's Law. However, today, according to reports from Boston Consulting Group and the Semiconductor Industry Association (SIA), the United States only accounts for 12% of the global semiconductor manufacturing share.

From almost monopolizing the entire market to only one-tenth of it, the decline of the US semiconductor manufacturing industry is obvious. The main reasons are firstly cost and secondly the industry ecology, including upstream and downstream supply chains.

Compared with other countries, especially Asia, the cost of semiconductor manufacturing in the United States is significantly higher. These costs come from the high construction costs in the United States on the one hand, and the talent costs in the United States on the other. Although the United States can provide the most cutting-edge technical talent reserves, it is insufficient for the engineering talent needed by the semiconductor manufacturing industry. According to the industry report of Eightfold AI in the United States, the current talent gap in the semiconductor manufacturing industry in the United States is 70,000 to 90,000, a proportion of up to 50%. In other words, if you need to find people immediately at the moment, you will need to pay a higher cost. This was also confirmed at the TSMC press conference last month. TSMC's Liu Deyin said that it was much more difficult to recruit talents for the construction of a factory in Arizona, USA than expected. Of course, the cost problem can be partially solved in the form of government subsidies. This is also the reason why many semiconductor giants in the United States want to increase investment after the government announced a large subsidy plan, but once such subsidies cannot be realized, the cost will be an important reason for the lack of competitiveness of the US semiconductor manufacturing industry.

TSMC Chairman Liu Deyin discussed the difficulties encountered in building a factory in the United States at the annual shareholders meeting

In addition to cost, the overall supply chain is also a reason that limits semiconductor manufacturing in the United States. As the overall electronics industry supply chain leaves the United States and enters Asia, the upstream and downstream industrial chains that interface with semiconductor manufacturing are also moving further and further away from the United States. Generally speaking, it is most advantageous for the entire industrial chain to be close to each other, including chip manufacturing, packaging and testing, PCB board manufacturing, and the assembly of the entire electronic system. If some core links are far away from other upstream and downstream, it will bring challenges to the overall production, including transportation time and cost, coordination difficulty, etc. At this point, Asia (especially Shenzhen and Taiwan in China) has the most complete overall supply chain, while the United States' supply chain is relatively incomplete. Even if chips are produced in the United States, other links such as PCB boards and assembly must be completed outside the United States, which also brings challenges to the US semiconductor manufacturing industry.

In the long run, unless the United States can continue to invest heavily in the semiconductor industry, the decline of the semiconductor manufacturing industry in the United States will still be a high probability event. Whether the CHIPS Act funding can be passed will be an important signal of whether the United States is determined to invest in the semiconductor industry. If the bill is shelved, the confidence of the semiconductor industry in the United States will continue to decline, and it is estimated that it will further move out of the United States. If the United States uses the CHIPS Act as a symbol to start investing heavily in the semiconductor industry for a long time (the possibility is not great), then the United States may still be able to slowly cultivate a set of semiconductor manufacturing and related industrial chains that can meet its core needs in the United States in ten years.

For the United States, its core demand may not be for the U.S. semiconductor manufacturing industry to return to its leading position in the world, but more to ensure that the needs of sensitive areas such as U.S. defense can be met, and that the United States does not fall behind in core technologies such as next-generation semiconductor manufacturing. Therefore, we believe that from the perspective of semiconductor manufacturing market share, regardless of whether the USICA bill can be passed, the U.S. semiconductor industry will not grow much in the next few years, and may even decline further.

Asia's leadership in the semiconductor industry will not be challenged by the United States, especially in the advanced semiconductor manufacturing and packaging and testing industries. Asian companies have the world's most advanced technology (TSMC and Samsung), and Asia also has the most complete supply chain. In the future, Asia will continue to be the global semiconductor and electronics industry manufacturing integration base.

In addition to Asia, Europe is also ambitious in the semiconductor field. The EU Chips Act released by the European Union in February this year is expected to allocate $43 billion to the semiconductor industry, which has attracted many giant companies, including Intel and GlobalFoundries, to further increase their investment in Europe (for example, Intel announced that it will invest $80 billion to build a factory in Europe, and GlobalFounreis just announced last week that it will cooperate with ST Microelectronics to invest billions of dollars to build a new 300mm in France to produce 22nm FD-SOI Fab. According to the EU Chips Act, Europe hopes to have a 20% share of the global semiconductor manufacturing market by 2030. Whether this goal can be achieved is still unknown, but Europe is indeed likely to have a certain degree of competitiveness in some differentiated fields (such as automotive electronics, analog and mixed-signal processes, rather than the most advanced semiconductor processes) to meet local needs.

*Disclaimer: This article is originally written by the author. The content of the article is the author's personal opinion. Semiconductor Industry Observer reprints it only to convey a different point of view. It does not mean that Semiconductor Industry Observer agrees or supports this point of view. If you have any objections, please contact Semiconductor Industry Observer.

Today is the 3104th content shared by "Semiconductor Industry Observer" for you, welcome to follow.