What should the UK do about semiconductors? Part 1: the UK’s place in the semiconductor world

The UK government is currently in the process of writing a new strategy for semiconductors. This is the first of a series of three blogposts setting out the context for this strategy.

In this first part, I discuss the new global environment, in which a tenser geopolitical situation has revived a policy climate around the world which is much more favourable to large scale government interventions in the industry. I’ll sketch the global state of the semiconductor industry and try to quantify the UK’s position in the semiconductor world.

In the second part, I’ll discuss the past and future of semiconductors, mentioning some of the important past interventions by governments around the world that have shaped the current situation, and I’ll speculate on where the industry might be going in the future.

Finally, in the third part, I’ll ask where this leaves the UK, and speculate on what its semiconductor strategy might seek to achieve.

As recent events have shown, the semiconductor industry is one of the most strategically important industries in the world, so it’s going to be very important for the UK government to get its strategy right. But there are more general principles at stake. We’re at a moment when a worldwide consensus behind the ideas of free trade and laissez-faire economics is being rapidly replaced in the major economies of the world by much more interventionist, and assertively nationalist, industrial policies. This isn’t comfortable territory for the British state, so how it responds to this test case will be very telling.

War, Semiconductors and the CHIPS act

It’s been reported that Russia has been dismantling washing machines to extract their integrated circuits, for use in missiles. True or not, this story illustrates two important features of the modern world. Integrated circuits – silicon chips – are now ubiquitous and indispensable for modern living – they’re not just to be found in computers and mobile phones; they’re in automobiles, consumer durables, even toys. And modern precision-guided weapon systems depend on them, so with a European war entering its second year, their strategic importance couldn’t be more obvious.

If demand for integrated circuits and other semiconductors is ubiquitous, we’ve also been reminded that their supply isn’t secure. The pandemic led to severe supply chain disruptions, in turn leading to major losses of production in the global automobile industry. The manufacture of the most technically advanced integrated circuits is concentrated in a single company – TSMC – located in the contested territory of Taiwan. This dependence means that, if the People’s Republic of China invades Taiwan, the consequences to the world economy would be disastrous.

This is the context for the USA’s CHIPS and Science Act – a hugely significant, and expensive, government intervention to rebuild the USA’s manufacturing capacity in the most advanced semiconductors. Underlying this is a serious attempt to restore its own technological supremacy – and specifically, to maintain its technological superiority over China.

This is the return, at scale, of industrial strategy. The primary driving force, as it was in the 1950’s and 60’s, is geopolitics, but the economic and political dimensions are important too, with an emphasis on restoring manufacturing – and the good jobs it provides – to communities that have suffered from deindustrialisation. The Act provides for expenditures, over five years, of $39 billion on incentives to return more semiconductor manufacturing to the USA, $13.2 billion for additional research and development, and $10 billion to create regional innovation hubs in economically lagging parts of the country.

It’s worth stressing what an ideological about-turn this represents. An economic advisor to the first President Bush reputedly said “Potato chips, computer chips, what’s the difference? A hundred dollars of one or a hundred dollars of the other is still a hundred dollars”. This is a marvellously succinct expression of the neoliberal argument against sector-based industrial strategy. It’s now clear how naive this view was. Crisps weren’t about to see the most rapid period of technological progress in history, propelling those countries like Taiwan and Korea that took advantage of this opportunity, from middle income economies, into the ranks of rich countries at the technological frontier. And Frito-Lay doesn’t make missiles.

The European Union has responded with its own European Chips Act. This includes an €11 billion “Chips for Europe Initiative”, together with further coordination of R&D and education and skills initiatives. Most significantly, it proposes a relaxation of state aid rules, allowing member states to directly subsidise new manufacturing facilities in Europe.

How should the UK respond to this new environment? The government is preparing a Semiconductor Strategy, but this has been repeatedly delayed.

The global semiconductor industry

What are the products of the global semiconductor industry? The most high profile are enormously complex integrated circuits that power our personal computers, gaming stations and mobile phones, as well as driving the giant server farms that underly cloud computing. The most important component of modern electronics is the transistor, a solid state switch. A few transistors can be combined to make a logic gate – the basic unit of a computer; the way this is done is described as “complementary metal oxide silicon” – hence CMOS. An integrated circuit combines a number of transistors on a single piece of silicon – a chip. Different designs of integrated circuits produce central processing units (CPUs), graphical processing units (GPUs), and solid state memory.

The more transistors the chip has, the more computing power or the bigger the memory, so the history of microelectronics is a story of miniaturisation, with each generation of chips having more transistors on a single integrated circuit, as expressed by Moore’s law. A modern CPU (such as Apple’s M1, made by TSMC) has 16 billion transistors, each of which has dimensions measured in nanometers. These are made by the most sophisticated and precise manufacturing processes in the world, through the successive deposition of layers of different materials, at each stage etching the layers with patterns that define the components.

Only three companies in the world have the capability to operate at this technological frontier: the USA’s Intel, Korea’s Samsung, and Taiwan’s TSMC. In recent years, progress at Intel has stumbled, and TSMC has taken a commanding lead for the manufacturing the highest performance integrated circuits. TSMC focuses purely on manufacturing, making integrated circuits to the designs of so-called fabless companies, such as Nvidia. Intel, on the other hand, designs its own chips and manufactures them.

The scale of capital investment required to make these advanced circuits is breathtaking. TSMC is reported to have invested $60 billion in its facilities to manufacture chips at the 3 nm and 5 nm nodes. TSMC has been incentivised by the US government to establish production in Arizona, at a cost of $40 bn. These huge capital sums reflect the high cost of the ultra-sophisticated, high precision equipment required to pattern these circuits on the nanoscale. The frontier processes rely on the extreme-UV lithography systems made by the Dutch company ASML, a single unit of which may cost $150 million. Other important centres of equipment production include Japan and the USA.

There is still substantial demand for less advanced integrated circuits, for applications in cars, consumer durables, industrial machinery, weapons systems and much else. In addition to the three industry leaders, companies like Global Foundries, STMicro and NXP operate manufacturing plants in the USA, Europe and Singapore. China’s leading semiconductor company, Semiconductor Manufacturing International Corporation, falls into this category, though it has aspirations to reach the technological frontier, and is supported in this goal by China’s government.

Not all semiconductors are silicon. Other materials – compound semiconductors, such as Gallium Arsenide, and Gallium Nitride – are particularly important for optoelectronics; the business of converting electricity to light and back again. These are the materials from which solid state lasers and light emitting diodes are made ; familiar in everyday life as scanners in supermarkets and low energy light bulbs, but no less importantly the technologies which make the internet possible, converting electronic signals into the optical pulses that transmit information at huge rates through optical fibres.

The primary driving force for innovation in semiconductors has been information and communication technology – the desire for more powerful computers and the higher rates of data transmission that make possible today’s internet. But information processing isn’t the only important use of semiconductors. In power electronics, the focus is on the switching, amplifying and transformation of the much higher currents needed to drive electric motors. These technologies are rapidly growing in importance; the transition to a net zero greenhouse gas energy economy is going to be driven by the replacement of internal combustion engines by electric motors. The growth of electrical vehicles, the growing importance of renewable energy and the need for energy storage, all will drive the need to efficiently handle and transform high power electricity using light and efficient solid state devices.

The UK’s place in the semiconductor world

The UK is not a big player in the global semiconductor industry. Its exports of integrated circuits, worth $1.63 bn, represent 0.24% of the world’s trade; insignificant compared to the world’s leaders, Taiwan, China and Korea, whose exports are worth $138 bn, $120 bn, and $89.1 bn respectively. Outside the Far East, the USA exports $44.2 bn; it’s this relatively weak position relative to the East Asian countries that has prompted the measures of the CHIPS Act. In Europe, the leading exporters are Germany and Ireland, at $12.8 bn, and $11.2 bn respectively.

As mentioned above, the manufacture of integrated circuits is hugely capital intensive, so it’s important to look at the suppliers of the equipment used to make chips. The export trade here is dominated by Japan, the Netherlands and the USA, worth $12 bn, $11.7 bn, and $10.7 bn respectively. The UK has 1.06% of the world market, with exports worth $497m.

One other important component of the supply chain for chip manufacture are the chemicals and materials needed. These include the silicon single crystals from which the wafers are made, amongst the purest substances ever made by man, a wide range of industrial gases and solvents and reagents, all supplied at very high purity grades, and highly optimised speciality chemicals – e.g. the materials that make up the photoresists. This sector is dominated by Japan, with exports worth $4.23 bn worth, representing 29.5% of the world trade. Here the UK exports $212 m, a 1.48% share of the world market.

It’s worth reflecting on these figures in the context of the UK’s overall trade position. The total value of its exports in 2020 were $700 bn, made up of $371 bn in products, and $329 bn in services, so these three semiconductor-related sectors amount to about 6.3% of its total product exports. But as these figures emphasise, service sector exports are particularly important for the UK, and this bigger story is mirrored in the semiconductor sector.

The most significant semiconductor company in the UK doesn’t make any semiconductors – ARM designs chips, deriving its income from royalties and licensing fees for its intellectual property. Its revenues of $2.7 bn in 2021 would have made a significant contribution to the UK’s service exports (2020 UK service exports included $21.3 bn in royalties and license fees). Smaller companies, such as Imagination and Graphcore, are similarly focused on design rather than manufacturing.

In recent years, the question of ownership of ARM has achieved prominence. Originally a public company listed on the London Stock Exchange, ARM was acquired by the Japanese finance house SoftBank in 2016. A proposed sale to the US firm Nvidia collapsed last year after concerns from regulators in the UK, the USA and the EU that the acquisition would seriously reduce competion. SoftBank still remains keen to sell the company, so the future ownership and control of ARM remains in question.

Sources

All trade figures 2020 numbers, from the Observatory of Economic Complexity.

Up next: What should the UK do about semiconductors? Part 2: the past and future of the global semiconductor industry