Taiwan and China’s quest to catch up on chips

See also: CHIPS Act will transform the semiconductor sector

Semiconductors or microchips are the new oil — a scarce resource without which the world, as we know it, will stop functioning. Everything from our smartphones, cars, refrigerators, TVs and microwave ovens to trains, planes, traffic lights and sophisticated military equipment need those chips. To maintain its military superiority and economic prosperity, America believes it must pull way ahead of China in the design and manufacturing of the most sophisticated semiconductors. “Semiconductors are the ground zero of our tech competition with China,” US Deputy Secretary of Defence Kathleen Hicks said last week.

For its part, China believes it needs to catch up with the US fast or it could forever be relegated to second-tier superpower status like Russia.

Made in China 2025

See also: Testing for Resource library

China’s chip ambitions aren’t new. In 2015, Prime Minister Li Keqiang unveiled “Made in China 2025”, a 10-year initiative to modernise the world’s second-largest economy’s industrial capability focusing heavily on intelligent manufacturing in 10 strategic sectors to secure China’s position as a global powerhouse in high-tech industries such as robotics, aviation, new energy vehicles or EVs, and semiconductors.

In 2017, Beijing unveiled its National ICT Plan, to promote information and communications technology. A key objective of the plan was for China, which spends more money every year importing chips than it spends importing crude oil, to be self-sufficient in chips. The goal was for China to produce 70% of the semiconductors it consumes by 2025. Currently, China manufactures just over 18% of the chips it consumes. Analysts expect that figure to rise to around 24% by 2025, way short of its audacious 70% goal.

In late 2019, Beijing quietly raised the target to produce 80% of the semiconductors it consumes by 2030. It won’t even be making 20% of its own chips — or a quarter of the way to its ambitious goal — by the end of 2023, with just seven years to go.

See also: Testing for Resource library

How did China fall so far behind its publicly stated target? Blame former president Donald Trump and his trade sanctions aimed at preventing Beijing from catching up with America on sensitive technologies like semiconductors. Washington’s efforts to slow the development of China’s domestic chip industry through export bans on US-origin technology have helped stymie China’s chipmaking ambitions. US exports of Chips to China have nearly doubled since 2015 when Beijing first unveiled its Made in China 2025 plan, while US exports of chip equipment to China are up nearly threefold. Rather than relying less on the US and other western nations on chip equipment, as well as chips themselves, China’s reliance on foreign chip technology has grown exponentially.

Beijing is working hard to achieve some form of technological self-sufficiency. China has earmarked over US$150 billion ($207.1 billion) by 2030 to jumpstart the production of the domestic semiconductor industry.

But China cannot catch up so easily. As I explained in last week’s column, any company or country that wants to make the most advanced chips in the world needs to buy stateof-art equipment from a Dutch company called ASML because it has a monopoly on extreme ultraviolet (EUV) lithography tools. The US has barred ASML from selling those tools to Chinese companies.

ASML charges up to US$175 million per tool. Demand for the tools is so high that there is a current backlog of over two years. Only three companies have placed orders for EUV lithography equipment — Taiwan Semiconductor Manufacturing Co (TSMC), Intel Corp and South Korea’s Samsung Electronics.

EUV tools help the big three chipmakers produce the most advanced process nodes. EUV machines are used to make more advanced and smaller chips that go into high-end devices such as top-of-the-line smartphones. TSMC and Samsung are already mass-producing 3-nanometre or 3nm nodes. Nanometres measure how many transistors can be built on a single chip. So a smaller 3nm chip which packs in far more transistors is far more efficient and powerful than a chip over nine times its size.

For its part, Intel has been struggling with 7nm nodes, though it expects to challenge TSMC and Samsung in three years when its two customised chip foundries in Arizona are up and running.

Meanwhile, China’s largest chipmaker Semiconductor Manufacturing International Corp (SMIC) began mass production of 14nm nodes around the time when TSMC was mass-producing 3nm nodes. That is how far behind China is right now on the most sophisticated chips.

Sink your teeth into in-depth insights from our contributors, and dive into financial and economic trends

Washington is now reportedly mulling restricting SMIC from producing logic chips using a 14nm-class fabrication process by preventing the export of sophisticated tools that help the largest Chinese chipmaker manufacture such chips. Two US companies, Applied Materials and Lam Research, provided SMIC with the tools to make chips using a 14nm process. SMIC could still make chips using existing imported tools, but it would be prevented from obtaining additional machines, components and software, which eventually will be a drag on capabilities to make chips using the 14nm process.

The crawl to close the gap

Because China has realised that it will not be able to close the gap any time in the foreseeable future due to export bans on sophisticated chip equipment, it has adopted a two-pronged approach to be a leader in semiconductors. It is strengthening domestic production of basic chips and at the same time adopting workaround solutions to make higher-performance chips that are assembled using relatively standard transistors.

China wants to optimise the mature 28nm chips — developed 10 years ago and currently widely used to make chips used in our electronic devices. The 28nm chip may be a mature technology, but it can be improved and would be enough to supply much of China’s most basic needs. At the same time, China will continue to work on its own homegrown technology to build smaller nodes. It may take China five to 10 years to close the gap but at least it can take the first few steps in its long march to self-sufficiency.

For now, it is a slow crawl to end reliance on Western chipmaking tools. Shanghai Micro Electronics Equipment (Group) Co (SMEE), China’s biggest semiconductor equipment maker, had promised to deliver 28nm lithography equipment to chipmaker SMIC by the end of 2021. It missed the deadline but expects to deliver the tools later this year. By the way, ASML already delivers lithography equipment capable of making chips with 1nm nodes.

Another chipmaking area China is focusing on is advanced packaging — combining larger node chips to deliver enhanced performance. The idea is to use disruptive technology and make larger node chips work similarly or perhaps even better without making the chips nodes smaller. Chinese chipmakers like SMIC are also trying different materials to deliver slightly better performance from mature node chips. When you can’t get the machine to make what you want, it is best to improvise and squeeze that last drop of juice to close the gap with competitors.

TSMC has 55% share of the global foundry market or chips made under contract for companies like Apple, Qualcomm, AMD and Nvidia as well as nearly 90% of all the world’s most advanced semiconductors. What does Beijing think of TSMC? And what happens to TSMC if there is a military conflict?

Many users of Sina Weibo, the Chinese equivalent of Twitter, have wondered aloud why China’s People’s Liberation Army (PLA) can’t just take over TSMC’s foundries in Hsinchu by force instead of pouring hundreds of billions into chip R&D in China over the next 10 years trying to play catch-up.

The problem is that technology is not just about factories or sophisticated tools and equipment. It is also about highly qualified engineers, expertise and software. Even if PLA were to march into Hsinchu and take over TSMC’s state-of-art foundries and fly in top-notch engineers from Shanghai and Beijing the next day, it is highly unlikely that TSMC plants will ever make anything at all. Without the software and expertise, expensive EUV tools will become massive pieces of junk.

Long before the attack begins, TSMC’s top engineers will flee. Even if China could capture a few of them and force them to run the equipment, they might not be able to do it on their own. Sophisticated chip manufacturing is a team effort and you need engineers who are experts in various parts of the complicated process to work together to run the whole thing smoothly.

If China were to invade Taiwan, the most-advanced chip factory in the world would be rendered “inoperable”, because it relies on global supply chains, TSMC chairman Mark Liu said in a CNN interview that was broadcast on the day Pelosi landed in Taipei. “Nobody can control TSMC by force,” he noted. “Because this is such a sophisticated manufacturing facility, it depends on real-time connection with the outside world, with Europe, Japan and the US, from materials to chemicals to spare parts to engineering software and diagnosis.”

Liu, a US citizen, compared a potential conflict in Taiwan with China to Russia’s invasion of Ukraine. While the two conflicts are very different, the economic impact on other countries would be similar, he said. “The Ukraine war is not good for any of the sides, it’s a lose-lose-lose scenario,” he told CNN. “War brings no winners, everybody’s a loser.”

The head of the world’s largest chipmaker noted that his company sells chips to consumer-facing Chinese companies that need its services and constant supply of advanced semiconductors. He also noted that the Chinese invasion of Taiwan would cause economic turmoil for China, Taiwan as well as Western nations that rely on TSMC’s chips. “How can we avoid war? How can we ensure that the engine of the world economy continues humming? Let’s have a fair competition,” he said.

If military intervention is not a viable option, why can’t China just reverse engineer the process of creating high-end semiconductors the way it has copied other technologies? For one thing, it’s not that simple. A single EUV tool that costs between US$125 million and US$175 million is built with components from 5,000 suppliers who provide 100,000 components. The tool weighs about 200 tons and is shipped in 40 containers on three large cargo planes. Moreover, operating and maintaining an EUV machine is almost as complicated as making one.

While Beijing will continue its war of nerves with Taiwan over Pelosi’s visit, it is more likely to gradually build its own expertise to make advanced chips than take over Hsinchu foundries that it knows it cannot operate without the software, know-how and access to the supply chain of components.

Assif Shameen is a technology and business writer based in North America