The global semiconductor shortage arose from a perfect storm of the Covid-19 pandemic and exponential demand fuelled by widespread digitalisation. With so many industry sectors, companies and even national economies dependent on a steady supply of semiconductors, policymakers and businesses are scrambling to find future safeguards against further disruption. This will inevitably mean a geographic re-evaluation of investment in semiconductor fabrication facilities (fabs). The question is, how is this likely to pan out?
The shortages have been attributed to bottlenecks in the global supply chain, but the underlying problem, put simply, is one of unprecedented demand, according to GlobalData analyst Anisha Bhatia, who says the shortage cannot just be attributed to economic uncertainties or weak supply chains. “It is because demand is booming,” she says.
Factors including longer manufacturing lead times as chips grow in complexity, accelerated digitalisation, rising consumer demand, increasingly busy shipping routes and geopolitical tensions that saw China stockpiling stores of chips, meant that the supply chain was stretched even before the global pandemic hit, says Bhatia.
Added to which, consolidation within the industry has meant that only a handful of companies run fabs. Many chip companies now focus instead on design, leaving the production of their designs to these large manufacturers.
Taiwan's semiconductor reliance is biggest global risk
In terms of geography, much of the focus for the shortages has been on the largest generalist foundry business, Taiwan’s Taiwan Semiconductor Manufacturing Company (TSMC), which has almost all its fabs located in Taiwan. Taiwan’s TMSC and South Korea's Samsung have a monopoly on the five-nanometre chips – the smallest chip in circulation used by companies including Apple, Qualcomm and Huawei.
TSMC leads the world in the output of foundry chips used by other companies. GlobalData principal analyst Emir Halilovic says this is clearly cause for concern. “TSMC and its clients are naturally aware of this, and they are taking steps to diversify geographically,” he says. For example, the company is in the process of building a new $12bn fab in Phoenix, Arizona, due for completion in 2024. The new facility is part of a $100bn investment announced in April to expand the company's manufacturing capacity.
However, fabs require huge investment and take a minimum of two years to build, providing no short-term solution. TSMC chief executive CC Wei told investors in May that the company would see the "supply chain tightness release a little bit" by 2023.
Asian governments tackle semiconductor shortage
Asian governments, as well as companies, are also focused on bolstering capacity. In May, the South Korean government announced the K-Semiconductor Strategy, which will see $450bn invested by 2030. The package includes tax breaks and incentives for chip businesses, training for 36,000 new staff, and $1.3bn invested in semiconductor R&D.
Although Taiwan leads in terms of manufacturing volume of semiconductors, South Korea leads in memory chips with a 65% global share, largely due to Samsung, according to research analyst Forrester. Samsung will invest $151bn in chip foundries over the next decade, an increase of $33.6bn on its existing expansion plans with competing South Korean company SK Hynix investing $97bn to upgrade existing facilities as well as $106bn on four new chip foundries.
The Japanese government has also pledged to fund half of a $348bn project with TSMC to build a new production facility near Tokyo due for completion in 2022. This follows TSMC’s announcement earlier in the year that it would invest about $178m in an advanced material R&D centre in the country.
Asia ahead on semiconductor talent pool
Asian governments have been offering companies subsidies and tax relief for the building of new fabs for a long while, according to Bhatia. “EU and US governments are just waking up to the fact, so I think we will see a lot more from them now,” she says. China and Singapore, in particular, have been fostering collaboration between universities and company R&D functions. “The US and the EU should be doing more of this by using the best university talent and expertise to upgrade private sector technology,” says Bhatia, adding: “That is why Asia is where it is in the manufacture of [semiconductor] technology.”
We see a crucial need for stronger education programmes as well as training and reskilling efforts to provide the highly skilled workforce needed to meet future demands. Ajit Manocha, SEMI
It is a concern echoed by industry body SEMI. President and CEO Ajit Manocha says the talent gap must be addressed to avoid future chip shortages. SEMI is partnering with governments and member companies worldwide on workforce development programmes. “We see a crucial need for stronger education programmes as well as training and reskilling efforts to provide the highly skilled workforce needed to meet future demands,” says Manocha.
According to Forrester, 79% of all the world’s chips were manufactured in Asia in 2019. Opening new fabs is one way of eliminating supply chain choke points, “preferably in areas less exposed to natural and political risks than East Asia”, says Halilovic.
Manocha adds: “While the majority of investments are in Asia, recent government incentives in the US and Europe are drawing attention but have to be finalised in order to possibly spark a geographical shift.”
In June, the US Senate passed the US Innovation and Competition Act, which included a proposal for $52bn in funding for the semiconductor industry, $39bn of which is to be spent on production and R&D incentives, and $10.5bn on a National Semiconductor Technology Centre. In May, the European Commission announced a joint declaration by 22 EU member states committing to bolstering Europe’s capacity for processors and semiconductor technologies.
However, despite US and European policymakers starting to foster home market ecosystems, Halilovic points out that geographical diversification cannot eliminate all potential negative impacts. “Foundry businesses are for-profit, they can’t afford to have ‘spare’ capacity lying around," he says. "Even if there are geographically dispersed foundry capacities, a crisis impacting any country or region with a lot of foundries will inevitably cause supply shortage.”
Will global cooperation resolve semiconductor shortage?
Rising global demand for semiconductors means that it is unlikely that an increase in regional investment will isolate risk for any one location or business. However, despite a common need for semiconductors, the industry is not homogenous and includes many different company and region-specific strengths.
Foundry businesses are for-profit, they can’t afford to have ‘spare’ capacity lying around. Emir Halilovic, GlobalData
For example, Intel dominates in terms of desktop and laptop central processing units, Qualcomm leads in smartphone chips, TSMC is the top manufacturer of ten-nanometre chips, South Korea’s Samsung leads the memory market, the US’s Nvidia dominates graphic cards, Japanese and Korean companies produce the majority of wafers, and the list goes on. While semiconductor companies and geographic hubs of expertise may excel in certain areas of design and manufacture, few have capabilities across the board.
Specialisation brings competitive advantages, but it also means semiconductor companies are highly interdependent. According to consulting firm McKinsey, no local market or company has all the capabilities required for end-to-end semiconductor design and manufacturing.
Indeed, Halilovic says a crucial root cause of the current shortages appears to be unequal technology evolution across the whole ecosystem. This means that not only do some geographies end up specialising in certain technologies, but there is also a discrepancy between some countries using older processes for chip manufacture. This technology disparity between global locations risks major disruption to the supply chain, as happened during the Covid-19 crisis.
This complex global supply chain of different components and chips used across many sectors means that the current semiconductor supply shortage is not directly connected to any single factor within the global market. Every geography, company and region has to take action to get things back on track, according to Bhatia. There are no winning or losing locations in such a globally interdependent industry. Strong growth projections for chips over the next decade fuelled by emerging technologies such as autonomous vehicles, AI, high-performance computing and 5G means that the problem needs long-term and sustainable solutions. “It doesn’t make sense to just increase capacity in particular geographies. It has to be on a global scale,” says Bhatia.