The resilience myth: fatal flaws in the push to secure chip supply chains
In the sweltering Asia summertime of mid-June, Taiwan Semiconductor Manufacturing Co urgently dispatched a team to Japan to visit some of the company’s equipment suppliers. Why, it wanted to know, were these companies saying they could not deliver vital machines on time? TSMC is the world’s largest chip manufacturer, and its suppliers had always bent over backward to provide what the powerful company was demanding but, for the first time, it was being met with apologetic messages.
The situation was highly sensitive. TSMC is in the midst of a $100bn expansion, spurred on by governments in the wake of last year’s alarming shortages of crucial chips. But the Taiwanese giant has found its own supply chains to be plagued by bottlenecks, affecting items that range from lenses so precise they could focus a laser beam on a pingpong ball on the moon, to apparently mundane valves and tubes.
The June mission followed on the heels of a similar trip by the company’s supply chain management chief, JK Lin, and a task force to the US in March, to investigate why the chipmaking machines TSMC ordered there were taking up to 18 months to turn up.
In Japan, suppliers including Tokyo Electron, the country’s largest chipmaking equipment manufacturer, and Screen Semiconductor Solutions told TSMC they might miss even the elongated delivery times they have promised, sources familiar with the tricky meetings told Nikkei Asia.
Screen — one of the few companies in the world making the chemical cleaning machines that are vital in chipmaking plants — reeled off a list of obscure components that it was having trouble securing from its own supply chain. Valves, tubes, pumps and containers made of special plastics — all are in short supply.
The problems are cascading from supplier to supplier and making it hard to resolve the global shortage of chips, the hearts and brains that power electronic devices from PCs and smartphones to automobiles.
The difficulties underscore a series of inconvenient truths, not just for TSMC and its rivals and suppliers, but for policymakers around the world. Amid US-China trade tensions and pandemic disruptions, governments in China, the US, Europe and elsewhere have determined to “onshore” semiconductor manufacturing. So-called supply chain resilience has become a central aim of policy. But such resilience is a myth.
These new national efforts are backed by huge subsidies and state-backed investments. The US Senate at the end of July approved the $52bn CHIPS Act. Japan’s government will back TSMC to the tune of ¥476bn ($3.5bn) to build a factory there for the first time.
The trouble is these efforts touch only the visible end of the semiconductor supply chain. Behind chip production sits a network supplying equipment and other items encompassing hundreds of raw materials, chemicals, consumable parts, gases and metals without which the bogglingly precise process of chipmaking could not function. China is directing a combined Rmb1.5tn ($221bn) of public and private investments to replicate a chip supply chain within its own borders, with modest results to date.
While a globalised semiconductor industry used to run smoothly across dozens of countries, the effort to replicate this architecture inside single countries or regions has revealed and exacerbated bottlenecks in the supply chain, according to Nikkei Asia’s investigations and interviews with more than two dozen senior industry executives from the major chip economies of the US, EU, Taiwan and Japan over the past five months. At the same time, there are questions over the long-term wisdom of the policy, and fears about whether, if they can be gotten up and running, many of these factories might ultimately sit idle.
JT Hsu, head of semiconductors and materials at Boston Consulting Group, said even a goal of reaching 70 to 80 per cent self-reliance is “extremely tough . . . It could be extremely challenging for any country or region to get all the fronts covered.”
This article is from Nikkei Asia, a global publication with a uniquely Asian perspective on politics, the economy, business and international affairs. Our own correspondents and outside commentators from around the world share their views on Asia, while our Asia300 section provides in-depth coverage of 300 of the biggest and fastest-growing listed companies from 11 economies outside Japan.
“It’s not only the [factories] that manufacture the chips but it’s everything that goes in there,” said Jens Liebermann, vice-president of semiconductor materials at the electronic materials business unit of BASF, the German chemical group. “All the materials, chemicals, gases and their raw materials. All have to be there. It comes down to, where is the source, where is the raw material, where is the manufacturing, and who can handle the logistics?”
Morris Chang, an elder statesman of the semiconductor industry who founded and formerly chaired TSMC, put it most bluntly in remarks addressed to the US.
“If you want to re-establish a complete semiconductor supply chain in the US, you will not find it as a possible task,” he said at an industry forum last year. “Even after you spend hundreds of billions of dollars, you will still find the supply chain to be incomplete, and you will find that it will be very high cost, much higher cost than what you currently have.”
Despite how insignificant they might sound, those valves, tubes, pipes, pumps and containers are a case study in complexity — and they are driving executives mad.
“I am not kidding! We are still receiving valves and tubes that we ordered more than a year ago,” one executive with a Taiwanese supplier to TSMC told Nikkei. “When opening the box, we are often very shocked. The box might contain only 10 pieces out of a 100-piece order.”
With only a handful of specialist suppliers able to meet anti-contamination standards and deal with the red tape of manufacturing items that also have potential military use, it has been no easy task to increase capacity, especially with limited supplies of the raw materials behind them.
These items are made of special plastics called fluoropolymers and are indispensable to the handling of corrosive chemicals and ultrapurified water that flows in all chip manufacturing facilities and chipmaking machines, where standards keep going up and up.
The most advanced chips, those used to build the latest iPhone and MacBook processors, for example, are now at the 5-nm level. Nanometre size refers to the line width between transistors on a chip. A nanometre is roughly 1/100,000 of the thickness of a piece of paper or human hair. The smaller the nanometre size, the more cutting-edge and powerful the chips are, and thus more challenging to develop and produce. In turn, chipmakers need to place billions of transistors on a chip. The tolerance for a defect or microcontamination is extremely low.
“The size of a Covid virus is about 100 nm,” Kevin Gorman, senior vice-president of integrated supply chain transformation with Merck Electronics of Germany, told Nikkei. “You can then see how refined the chip manufacturing work is and why all the materials are critical.”
When it comes to semiconductor-grade valves and tubes for handling chemicals, it is crucial they do not become a source of contamination. Only a few suppliers worldwide have the capability to reach the exacting requirements, according to Nikkei Asia analysis. CKD and Advance Electric of Japan and Entegris of the US, are qualified suppliers of valves; Iwaki of Japan is the dominant supplier for chemical-handling pumps; industry sources referred to Agru of Austria and Georg Fischer of Switzerland as essential providers of the critical piping systems for chip plants.
The Wassenaar Arrangement, a multinational agreement signed by more than 40 nations to avoid such components being shipped to rogue states for military use, adds red tape that provides another barrier to new entrants.
Follow the supply chain upstream, and further chokepoints emerge with regard to the fluoropolymers from which these components are made. One such material, known as PFA, is only supplied by Chemours of the US and Daikin Industries of Japan. It requires extensive knowhow to process, and no competitors are on the horizon.
Other key fluoropolymer material makers include Solvay of Belgium, 3M of the US, Gujarat Fluorochemicals of India and HaloPolymer of Russia. But not all of them are qualified to build semiconductor-grade materials and they must supply to a wide range of other industries beyond the tech sector. Sources from Russia have dropped away due to the disruption and sanctions caused by its war in Ukraine.
Hsu Chun-yuan, chief business development officer of United Integrated Services, a leading cleanroom builder for TSMC and rival chipmaker Micron Technology, told Nikkei that “sources of fluoropolymers are constrained” and there have been “demand hikes from both the chip and battery industries, driven by the electric vehicle boom”.
And further upstream still? Fluoropolymers are processed from fluorspar, also known as fluorite, a mineral of which China controls nearly 60 per cent of the global production output, according to data from market research company IndexBox. China has long identified fluorspar as a strategic resource and back in the late 1990s limited exports due to its importance to industries from agriculture, electronics and pharmaceuticals to aviation, space and defence. The mineral is often labelled as a “semi-rare earth”.
According to IndexBox, Mexico is the second-largest producer of fluorspar with about 10.8 per cent of the market last year, followed by Mongolia and South Africa. In Europe, Bulgaria and Spain together control some 5 per cent of the global market. In a supply chain review paper published by the White House in 2021, the US flagged the risks of critical materials subject to foreign domination and identified fluorspar as one in a list of “shortfall strategic and critical materials”. The report did not point out its deep link with the chipmaking industry. It said increasing sources of critical minerals, strengthening stockpiles, and ramping up North American manufacturing, processing, and recycling capacity could result in fewer disruptions during “future worldwide crises”.
Similar issues arise in the handling of gases such as neon, used in lithography, and C4F6, a fluorine gas used in etching. Both count either Ukraine or Russia as a major source of supply, which has been disrupted by the war. The equipment for moving them around is also highly specialised.
Only a handful of companies — including Rotarex of Luxembourg and BBB Neriki Valve and Hamai Industries of Japan — are qualified to supply the ultra high purity valves for the gas cylinders that the semiconductor industry uses, Nikkei Asia supply chain checks show. Rotarex controls close to 80 per cent of the market and only produces these specific items in Luxembourg.
The valves, built with stainless steel and other alloys, must endure extensive verification processes and need to be government certified because of the dangers of leaks and explosions. It would take “10 to 20 years” for a new entrant to meet the standards and tests of different government authorities for certification, some industry executives told Nikkei.
Trade tension, pandemic and war
The call for chip supply chain resilience emerged amid the US-China tech war when former US president Donald Trump’s administration clamped down on Chinese tech champion Huawei Technologies in 2019 and blocked its use of American technologies, especially chips, citing national security. The drastic move spurred an aggressive nationwide Chinese campaign across sectors to cut dependence on the US and build a secure, self-controllable supply chain.
The self-sufficiency movement evolved into a global campaign in late 2020, as unprecedented chip shortages stalled car production and hurt a wide range of industries, crimping global economic growth and threatening jobs. The US Department of Commerce said the shortages wiped an estimated $240bn off the country’s gross domestic product in 2021. The automobile industry alone made 7.7mn fewer cars than the year before.
The Ukraine war has further amplified demands for supply chain security. The war drove up prices of energy, metals, chemicals and crucial gases that many chip-related suppliers needed. It also increased the sense of urgency.
For most major economies, chips are essential for building everything from computers and data centres to appliances and cars. They are central to the battle for supremacy in space, science, artificial intelligence and EVs, and will be crucial to the military and defence equipment of the future. Advanced chips are integral to an array of critical national security capabilities “including sophisticated weapons systems such as the Javelin antitank missiles the US is supplying to Ukraine to defend itself against Putin’s invasion”, the US Department of Commerce pointed out in a recent report.
Governments so far have promised to pour more than $100bn into subsidising the building of local chip supply chains. As well as the US CHIPS Act, the EU adopted the €45bn ($46bn) European Chips Act, Japan had budgeted ¥600bn and India set up a $30bn funding programme for semiconductors and other tech sectors.
Major chipmakers from Intel, Micron and Texas Instruments in the US to TSMC and South Korea’s Samsung Electronics have separately announced more than $650bn in investments. These include several outside their home bases. TSMC is building in the US and Japan, Intel plans to expand in Europe and south-east Asia, and Samsung has construction plans in the US. According to SEMI’s estimate, some 91 new chip plants are set to go online worldwide from 2020 through 2024.
When the European Chips Act was enacted earlier this year, European Commission president Ursula von der Leyen acknowledged that “no country — and even no continent — can be entirely self-sufficient”. The hope is that parts of the supply chain that cannot be brought onshore will at least run through friendly nations.
“Europe will always work to keep global markets open and to keep them connected. This is in the world’s interest; it is in our own interest, too,” she said. “Europe will build partnerships on chips with like-minded partners, for example the United States or for example Japan. It is about balanced interdependencies and it is about reliability.”
US Treasury secretary Janet Yellen has floated “friendshoring” as a compromise concept. “We cannot allow countries to use their market position in key raw materials, technologies, or products to have the power to disrupt our economy or exercise unwanted geopolitical leverage,” she said in April. “Let’s build on and deepen economic integration and the efficiencies it brings, on terms that work better for American workers. And let’s do it with the countries we know we can count on.”
Russia’s fall from western favour demonstrates that alliances can shift over time and spats can emerge even between nations ostensibly committed to free trade.
Japan limited the export of photoresists, a crucial chipmaking chemical dominated by Japanese suppliers, to South Korea during a Tokyo-Seoul trade war in 2019.
An assessment by BCG suggests there are at least 50 chokepoints in the semiconductor supply chain across design tools, manufacturing, packaging, materials and equipment. These points are defined as areas where 65 per cent or more of a particular item is concentrated in a single country or region.
The US dominates chip design tools and at least 23 types of essential equipment, it found. Japan is a leader in the production and critical formulation of critical materials that include wafers as well as photoresists. Europe is the leader in industrial gas.
The extreme ultraviolet (EUV) lithography machine exclusively built by ASML of the Netherlands provides a prime example of how difficult it is to switch a component in the chip supply chain. Sometimes it is just impossible to find alternatives.
The EUV machine is indispensable in the production of cutting-edge chips of 7 nm and below, helping project the complicated patterns of integrated circuits on a microscale. Production delays are hampering the ability to add new capacity, lengthening the current chip crunch and setting back the introduction of more cutting-edge chips.
ASML has extended the waiting time for several models to two years due to constraints on vital parts including optical mirrors and lenses, people familiar with the matter told Nikkei. A company spokesperson acknowledged some delays and said constraints on the industry were “very diverse and across multiple tier suppliers”.
Creating EUV light inside a vacuum chamber within a machine is exceptionally challenging, relying on Germany’s Trumpf for a powerful laser source and another German partner, the optics specialist Zeiss Group, for a system of mirrors to reflect and direct the light.
Since even the smallest irregularities cause aberrations, Zeiss boasts that its product is the world’s “most precise” mirror. “If one of these EUV mirrors were to redirect a laser beam and aim it at the moon, it would be able to hit a pingpong ball on the moon’s surface,” CEO Andreas Pecher told Nikkei. Zeiss and ASML have been working together for nearly 30 years.
Even if ASML wants to strengthen its own supply chain resilience and looks for other optical partners, it will require at least five to 10 years of co-development work before getting initial results, several executives told Nikkei.
“Actually it’s almost not replaceable in the many years to come,” Nikkei heard from one executive at a Japanese lens maker.
There is almost no part of the chipmaking process that does not require deep specialisation and no part of the supply chain that can be simply and quickly duplicated.
Chemicals and solvents used in chip plants need to reach the so-called part-per-trillion (PPT) grade — one particle to 1tn drops. Gases need to reach a purity of up to 99.9999 per cent — the so-called 6N — when it comes to cutting-edge chip production. For silicon wafers, the basic substrate materials that chips are fabricated on, all need to be as pure as 9N, or 99.9999999 per cent, an executive with the chip material distributor Wah Lee Industrial told Nikkei.
“If you want a resilient chip supply chain, you not only need chip plants, you also need a whole string of suppliers from critical chemicals and precision components all coming along,” said an executive at Japan’s Daikin. “Building a semiconductor plant takes several years, but building chemical plants will take even longer given the extensive environmental assessments and regulations for handling chemicals.”
The long road to onshoring
China’s efforts demonstrate that the practical difficulty of building a chip supply chain cannot be overcome by throwing billions of dollars into the effort. As early as 2014, Beijing launched the first phase of the China Integrated Circuit Industry Investment Fund, nicknamed the Big Fund, with Rmb138.7bn. Another Rmb204bn followed in 2019. The first national seed fund stimulated more than Rmb500bn of investment from the private sector and local governments; the second phase of the fund is expected to encourage a further Rmb1tn.
China indeed increased local chip production — to 16.7 per cent of its domestic needs in 2021 from 12.7 per cent a decade before, IC Insights data show.
The mathematical implication of having many countries creating new onshore chip supply chains is that capacity is going to be much greater than the world as a whole actually needs.
The industry has indicated that these are often noneconomic investment plans by saying that, in many cases, factories will only be built if they are heavily subsidised. With consumer spending on electronics apparently slowing sharply and recession talk in the air, the outlook for actual chip demand, at least in the short term, is suddenly uncertain.
Gorman of Merck Electronics acknowledged questions about whether local plants could reach economic scale, but said it still makes sense to localise if its key customers could together shoulder the risks.
“Keeping the supply line short is also better for our environment,” he told Nikkei. “Our customers . . . will favour a local supply over one that has to cross international borders.”
Building an onshore chip supply chain is a “very large-scale and long-term journey”, BASF’s Liebermann told Nikkei. “It will take a lot of time and a lot of costs and the cost will only be justified if the utilisation rates of those new plants are meeting the demand, and the demand is high enough.”
Most industry executives believe a long-term increase in chip demand is locked in, regardless of the current economic environment, as everyday items become more connected and complex and as cars go electric and, ultimately, autonomous. A semiconductor industry that had revenues of nearly $600bn in 2021 is widely projected to be at $1tn by 2030.
“If we really believe that the industry will be reaching $1tn . . . we should be able to have some level of regionalisation of the manufacturing and have the right leverage,” Bertrand Loy, CEO of Entegris, told Nikkei. “But we won’t be able to have manufacturing everywhere and get the right leverage. We are investing in some countries, some products, but not in all countries for all products because we cannot afford [to do] that.”
ASML believes regional investments “can coexist, if connected to a global ecosystem”, its spokesperson said. “Compartmentalisation leads to sub-optimisation, which leads to higher cost and slower innovation for consumers and companies and governments who rely on this innovation.”
‘No longer an era of free trade’
Simon HH Wu, president of San Fu Chemical, a Taiwanese chipmaking chemical supplier, reckons geopolitical conflicts and trade barriers are prevailing over globalisation, upon which the chip industry was built. “It’s no longer an era of free trade,” he told Nikkei, warning that policymakers and the industry should be under no illusions about the difficulties ahead.
“Any country that controls certain natural resources or key technologies would want to protect and leverage those resources for economic and political benefits,” Wu said. “What companies could do is to look for allies and partners to alleviate the potential disruptions.
“There’s always something you need to import and ship from another place, country or even continent. If you don’t have phosphate rock how do you produce chipmaking phosphoric acid? If you don’t have fluorspar, how do you produce fluoropolymers? At the end of the day, you can’t move all those mines and natural resources . . . next door.”
JT Hsu, the head of semiconductors and materials at BCG, said the chip crunch shows it is about time to build some “redundant” capacity to give the industry a buffer to absorb shocks. “However,” he said, “it’s nearly impossible and unrealistic that any country or region could reach a point of 100 per cent self-reliance, in terms of making everything about chips from the start to the end. That’s not possible now and that is not likely to be possible in the future.”
A version of this article was first published by Nikkei Asia on July 27. ©2022 Nikkei Inc. All rights reserved.