Implications of the 2021 Microchip Shortage
From the perspective of 2021, the underlying vulnerabilities of global supply chains are obvious. And there is no better or more far-reaching example of that than the semiconductor industry, where barriers to entry are notoriously high, technological competition is cut-throat, lead times are long, and the balance of power overwhelmingly tilts towards a few gargantuan companies.
Furthermore, other indispensable industries depend on semiconductors for their core products, including automobiles, personal electronics, and home appliances. So, with production backlogs forming as the global economy emerges from the COVID-19 pandemic, customers who depend on an uninterrupted supply of semiconductors are feeling the pain. Auto manufacturers, for example, could experience a $61 billion loss in revenue due to semiconductor supply constraints in 2021.
The problem has escalated to the point where governments are getting involved, not only to help alleviate near-term bottlenecks, but also to develop policies that protect the stability of semiconductor supply chains in an effort to avoid disruptions over the long run. But despite the headwinds for various industries the semiconductor shortage creates in the short-term, it is important not to lose sight of the long-term trends.
In many ways, semiconductor chips are to the fifth techno-economic revolution what steel and oil were to the third and fourth techno-economic revolutions. That is, a wide range of critical industries and products depend on semiconductors to enable innovative new features and drive them toward lower costs. In this segment, we focus on the reasons for the underlying shortages and the critical geopolitical and technological implications of ensuring secure semiconductor supply chains.
From a geopolitical standpoint, it's important to remember that the United States is the global leader in the field of chip design, with American companies accounting for 47% of global semiconductor sales in 2019. What allows it to maintain this competitive edge?
First there is America's unique ability to attract talented engineers from abroad. Second, America leads the world in R&D spending as a percent of sales in the semiconductor industry, at 16.4%. However, today's fabless companies, like Qualcomm and Nvidia, only engage in chip design and sales. They outsource the actual production of chips to other companies. So, while most cutting-edge chip designs come from the United States, the lion's share of manufacturing takes place in East Asia, particularly Taiwan and South Korea.
Unfortunately, a perfect storm of geopolitical ten fabless business model is why almost half of global sales are attributed to U.S. companies, despite only 12% of global manufacturing capacity being located in the U.S., while 43% of global manufacturing capacity is located in Taiwan and South Korea. And, of all the Taiwanese and South Korean semiconductor companies, TSMC and Samsung overwhelmingly dominate the field because they have the fabs capable of making the most advanced chips in the world.
Furthermore, as the current shortage has demonstrated, it is difficult for semiconductor fabs to adjust their output in response to external shocks. Building a new fab to meet growing demand is simply not a feasible solution for short-term disruptions because of the immense capital and lead-time required to do so.
Building a fab and bringing it up to full capacity can take anywhere from 24 to 42 months at a price tag of anywhere from $1.7 billion to $5.4 billion, depending on the type of chips manufactured. And those costs are only increasing as semiconductors continue to become smaller and more complex. Furthermore, fabs require lead times for orders and cannot easily open capacity at a moment's notice.
Prior to the pandemic, lead times were already long enough, but the past year of disruption has extended them. Between January and April 2021, lead times reportedly increased an average of 75%, with some customers seeing increases of up to 52 weeks. Businesses which require semiconductors in their products must balance their orders with just the right amount of production to avoid having too many, or too few semiconductors on hand.
2021 sions, inflexible supply chains, high barriers to entry, concentrations of power in a select handful of companies, and the once-in-a-century shock of COVID led to a supply-side logjam that is now impacting multiple aspects of the economy. Before COVID, geopolitical tensions were already setting the stage for the ongoing shortage.
The US-China trade war saw sanctions leveled against key Chinese tech firms like Huawei and ZTE, which were cut off from buying chips made with US technology. In response to these sanctions, 2019 saw Chinese tech companies like Huawei and Hikvision scrambling to stockpile chips. When COVID struck, factories around the world were left with no choice but to shut down because of stay-at-home orders.
Despite the lockdowns, semiconductor companies were given a little more leeway; for instance, at the height of the lockdown in Wuhan, YTMC and XMC could continue operations. But, faced with unusual changes in demand, semiconductor companies could not adjust their production fast enough. To make matters worse, a combination of natural disasters and fab incidents exacerbated the shortage.
In February, an unprecedented blizzard led to power outages across Texas that brought production at fabs owned by Samsung, Infineon and NXP Semiconductor to a grinding halt. In March, a fire resulted in production stopping for nearly a month at a Renesas factory in Ibaraki Prefecture, Japan. Furthermore, an ongoing drought in Taiwan threatens to impact operations of major fabs run by TSMC. On the demand side, the uncertainty of COVID caught both companies and consumers off guard.
With unprecedented lockdowns and travel restrictions in place, it was not entirely clear how demand for various products would be impacted. Automotive companies, many of which prefer not to stockpile inventory, anticipated a decline in automobile demand and adjusted their plans accordingly. Meanwhile, with people confined to their homes, demand for household electronics remained resilient. So, while automotive companies cut back on chip orders, consumer electronics companies continued to order chips.
Fast forward to September 2020. At that point, car sales rose close to their pre-pandemic levels, and carmakers needed to ramp up production. When carmakers got back in line for semiconductor orders, chipmakers already had a significant backlog of orders from other companies. The trouble was that semiconductors are necessary for the functioning of modern vehicles and therefore a shortage of them can easily threaten production.
In fact, modern vehicles utilize hundreds of semiconductors each - a number that is increasing as cars become ever more advanced. As a result of the shortage, semiconductors' average share of each car's price increased from 27% in 2010 to 40% in 2020, and a rise in electric and autonomous vehicles is expected to further increase that number. As of May 2020, the semiconductor shortage has escalated to the point where governments of major economies feel the need to take action and even worldwide PC shipments and vehicle sales both took a hit in 2020 and followed a similar trajectory afterwards.
In the United States, the Biden administration: Signed an executive order calling for a 100-day review of supply chains; Held a semiconductor summit at the White House in April; Is pushing so-called "Chips for America" legislation, which earmarks $50 billion for semiconductor R&D; and Agreed with Japan to cooperate on semiconductor development and supply chains.
In Japan, the Abe and Suga administrations: Dispatched a delegation in June to negotiate with and invite TSMC to build a fab in Japan; Asked Taiwanese manufacturers in March to cooperate in alternative production of chips; and Encouraged equipment makers to support Renesas in the wake of the Renesas factory fire.
Meanwhile, South Korea responded by: Exempting businesspeople procuring auto chips from Korea's 2-week quarantine; Offering increased vaccinations for key people in the auto chip supply chain; Holding a "Blue House summit" with companies in the automotive and semiconductor industries; and Dispatching government officials to Taiwan to negotiate for more secure chip supply. Where is all of this going?
Generally speaking, all the most important disruptive technologies moving up the adoption S-Curve depend on better, cheaper, and faster semiconductors. Whether it's electric & autonomous vehicles, artificial intelligence, industrial robots, or IoT devices, almost all disruptive technologies require on semiconductors in one way or another. The bottom line is that despite short-term problems in the semiconductor industry, long-term trends will structurally increase semiconductor demand over at least the next decade. Meanwhile, customers are going to do everything possible to ensure an adequate and reliable supply.
Given this trend, we offer the following forecasts for your consideration. First, over the next decade, countries will reshore semiconductor supply chains, where possible, to prevent future disruptions. In fact, 53% of semiconductor companies in a 2021 survey identified "territorialism" as the biggest industry issue. However, reshoring the semiconductor industry is difficult, and not all efforts to do so will succeed.
As a chart in the printable issue shows, the geographical centers of power in the semiconductor industry shifted dramatically over the past 30 years. Japan once dominated as its companies generated 49% of integrated circuit sales in 1990. But by 2017, that number dwindled down to 7%, with most of it going to Japan's rapidly growing neighbors.
In 2020, Japanese officials invited TSMC to build a facility in Japan, which ultimately resulted in TSMC deciding to raise $9bn for a Japanese subsidiary in 2021. However, it is not clear how determined the Japanese government is to pursue reshoring.
Second, China will strive to build fab and design capabilities, but it will continue to lag in an industry where being on the cutting-edge is critical. Chinese policymakers have been trying to boost the semiconductor industry for a decade now. In 2015, the State Council put forth a goal of 70% self-sufficiency in semiconductors by 2025.
Although China is still far off track from meeting any such goal, it has not become complacent. If anything, pressure from U.S. sanctions on companies like Huawei and SMIC has spurred China's policymakers to step up their efforts. It's 14th "five-year plan," released in March 2021, strongly emphasizes technological innovation, and also includes third-generation semiconductors on a list of technologies the government aims to support with scientific research programs.
Third, the United States will continue to be the leader in chip design, even as it gains share in manufacturing. On April 12th, the White House held a special CEO summit with companies that make semiconductors or are affected by the shortage. Participants emphasized the importance of boosting semiconductor manufacturing in America. Intel has already made the decision to spend $20 billion on new chip plants in Arizona and $3.5 billion to upgrade an existing plant in New Mexico. Other companies could decide to follow suit.
Fourth, over the coming decade, the EU will increase its share of worldwide semiconductor manufacturing. European Union officials are already formulating plans to restore some of the faded glory of the continent's semiconductor manufacturing industry. It seems today's chip shortage is a sign that it is time to restructure supply chains. To achieve that goal, the EU Commission put forth a plan that aims to bring EU chip production up to 20% of world supply by 2030.
Fifth, the reshoring of the semiconductor industry will bring about a decline in the "fabless" model as companies vertically integrate from chip design to manufacturing. Intel's recent decision to build new fabs on American soil shows that it is committed to growing its foundry business. Intel's CEO Pat Gelsinger confirmed this when he stated that it is time to reverse the trend of decline in US chip manufacturing.
If other firms like Nvidia or even major tech firms like Apple decide to follow suit and pursue vertical integration, it will take a lot of capital and talent to overcome Samsung and TSMC's dominant position in manufacturing. Not all efforts to re-shore the semiconductor industry will succeed, but the attempt alone will probably be enough to put pressure on the fabless, global market.
Sixth, regardless of where it occurs, enormous capital expenditures will be required to meet the surging demand for the latest chips. On its Q1 2021 earnings conference call, Taiwan Semiconductor Manufacturing Corporation (or TSMC) said its decision to ramp up capital expenditure to $30 billion in FY2021 was because of "multi-year structural megatrends associated with high-powered computing (or HPC) and 5G-related applications." And, given how automated semiconductor fabs already are, at Trends we expect that much of this capital spending will go into industrial robots. Semiconductor companies like TSMC and Samsung have already increased automation in recent years.
In 2020, TSMC developed the world's first automated wafer transportation system, which can safely carry wafers from warehouse docks to fabs and reduce manual weight handling by 95%. This automated system was rolled out in Q1 2020 at its newest plant, and TSMC plans to roll out the system to all of its 12-inch Gigafabs in Taiwan before the end of 2021. Meanwhile, Samsung has fully automated 12 tasks at one of its newer foundries and it is expected to carry those over to its newest foundry, still under construction.
Seventh, innovations in AI, 5G, vehicle technology, and the Internet of Things will lead to a continuing rise in demand for the latest semiconductors throughout the coming decade. Extracting insights from massive datasets with AI and machine learning requires more processing power than most traditional applications. This processing power depends heavily on advances in semiconductor technology.
Because of this, new semiconductors that specially cater to the needs of AI and machine learning have gained traction over the past few years. Field Programmable Gate Arrays and Application Specific Integrated Circuits are two examples of semiconductors that suit the needs of AI. And these are likely to be major drivers causing the size of the AI chip market to increase eight-fold from an estimated $10.14 billion in 2020 to $83.25 billion by 2027.
Similarly, plans to roll out 5G technology are already noticeably impacting demand for semiconductors. The higher data rates and lower latencies made possible by 5G will drive demand for advanced semiconductors, and the concurrent growth of the Internet of Things will reinforce that demand surge.
In the short run, the shortage of semiconductors will create headwinds for 5G, but the actions of chipmakers suggest they believe these headwinds will not last for too long, as evidenced by TSMC's decision to ramp up capital expenditures to $100 billion over the next three years. In that same vein, a 2021 KPMG survey showed that 53% of semiconductor companies believe 5G will become a significant driver of revenue growth within one to two years, and 19% believe it could happen in less than one year.
Automakers have borne the brunt of the shortage thus far, but the shortage may also reduce smartphone production by 5% in Q2 2021. In fact, the shortage contributed to both Apple and Samsung delaying the launch of new phones. Meanwhile, manufacturers of other products like TVs, video game consoles and even appliances are beginning to feel anxious about possible risks. The wide-reaching impact of the shortage across subindustries is a testament to how ubiquitous semiconductors are.
And, Eighth, the real winners from the semiconductor shortage and subsequent supply chain shakeup will be those companies that pursue innovation. There is plenty of room for new concepts as Moore's Law continues onward. Companies that reach for "More Than Moore" innovation, namely finding new methods of chip optimization that are more efficient than traditional architecture, will have a better chance at disrupting the status quo of the semiconductor industry.
To seize such opportunities Apple has developed the M1 system of chips for tablets and MacBooks, Amazon has developed the Graviton chip for servers, Google has developed the Tensor Processing Unit for neural network machine learning and Alibaba has developed the XuanTie 910 suite of IoT technology.