Domestic car chips, dangers and opportunities

Against the backdrop of technological competition among major countries, semiconductors have become the first industry to receive common attention.

In order to seize the commanding heights of future technological development, major markets such as Europe, the United States, Japan, and South Korea have introduced policies to support the development of local semiconductor industries since the beginning of this year. According to relevant statistics, the total capital investment in the global chip industry is expected to reach US$190 billion in 2022 alone.

Among them, the United States, while developing its own semiconductor industry, has even formulated a series of measures to restrict the development of semiconductor industries in other countries in order to ensure its absolute leadership in the semiconductor field.

According to the latest report from Reuters, the Biden administration plans to expand export restrictions on AI and chip manufacturing to China next month.

Considering the rapid development of intelligence and electrification, which has greatly increased the demand for semiconductors, including advanced semiconductors, driven by automobiles, it is not difficult to predict that this series of "chip restriction" measures will also restrict the development of China's automobile industry to a certain extent.

Chip competition may exacerbate chip shortages

From the signing of the "Chips and Science Act of 2022" to the successive issuance of the EDA ban and the high-end GPU sales ban, the United States is stepping up its efforts to restrict the development of China's chip industry, which are continuously extending to equipment, chip design, wafer foundry, chip supply, application and other links.

As one of the most important application areas of semiconductors, the automotive industry is destined to not be able to remain immune to this game.

At present, from domestic chip research and development, to the supply of automotive chips, to related technological innovations that rely on chips, all may be affected by these bans.

On August 9, the United States officially signed the "Chips and Science Act of 2022", which clearly requires that companies that receive financial subsidies should not expand advanced chip manufacturing in China or other relevant countries in the next 10 years.

A few days later, the U.S. Department of Commerce issued an interim final rule to impose export controls on ECAD software necessary for designing GAAFET (gate-all-around field-effect transistor) structure integrated circuits. It is reported that this process is currently mainly used for advanced processes such as 2nm and 3nm.

Prior to this, foreign media reported that the US Department of Commerce had also required all US chip production equipment manufacturers to obtain relevant licenses before they could export 14-nanometer and below chip production equipment to China.

It should be noted that the United States is the number one powerhouse in semiconductor equipment, and in the core links of chips, the most obvious domestic shortcomings are in the upstream raw materials, design tools and mid- and downstream chip manufacturing and packaging and testing links, which has made domestic semiconductors heavily dependent on imports. Especially for automotive semiconductors, the self-sufficiency rate is less than 5%, and in core technology fields such as automotive MCUs, the market share of local chip companies is even less than 1%.

Now the Chips and Science Act of 2022 requires chip companies to "invest in the United States, research and develop in the United States, and manufacture in the United States", and proposes a subsidy policy that excludes China. For chip companies with multinational businesses, they may have to face the dilemma of "choose one of the two" in the future. Under pressure, it is not ruled out that some multinational chip manufacturers will slow down their domestic capacity expansion plans.

Although the current demand for automotive-grade chips is mainly concentrated in mature process chips of 28 nanometers and above, according to the partial implementation strategy of the "2022 Chip and Technology Act" released by the U.S. Department of Commerce on September 6, the bill plans to provide up to $52.7 billion in government subsidies to the U.S. semiconductor industry. It is clearly stated that approximately $10 billion will be used to expand the manufacturing of mature process chips used in automobiles and other fields. It is not difficult to predict that this will have a certain impact on the supply of domestic automotive chips.

Recently, TSMC, Samsung, Intel, Micron Technology, SK Hynix, etc. have announced a series of expansion plans in the United States.

Samsung was previously exposed to have proposed a potential plan to invest $200 billion in 11 factories, two of which will be located in Austin, Texas, and the other nine will be located in Tyler, Texas. Intel has reached a cooperation with Canadian asset management company Brookfield Asset Management to jointly invest up to $30 billion for Intel's previously announced manufacturing expansion in Chandler, Arizona, USA. Micron Technology announced in early September that it plans to invest $15 billion by 2030 to build a new cutting-edge memory manufacturing plant in Boise, Idaho. It is reported that the plant will focus on providing memory chips for the automotive and data center industries in the domestic US market. SK Hynix plans to select a site in the United States to build an advanced chip packaging plant and break ground around the first quarter of next year.

Regardless of whether these companies' expansion plans are passive or active, they will affect the release of domestic production capacity to a certain extent. Moreover, in terms of wafer processing, TSMC itself occupies an absolute majority, with a market share of more than 50%. In contrast, SMIC's market share is only 5%.

In view of this, considering that the automotive industry is facing chip shortages and price increases due to the epidemic and the rapid development of smart electric vehicles in the past two years, the implementation of the chip bill may further aggravate these problems in the short term.

Image source: Nanochip Microelectronics

According to relevant analysis data, the demand for automotive chips brought about by electrification may be twice that of traditional fuel vehicles. As intelligence moves towards L4 and L5 levels of autonomous driving in the future, the new demand for semiconductors may be 8-10 times that of traditional non-intelligent vehicles. Taking a luxury car as an example, according to Wang Shengyang, founder, chairman and general manager of Nanochip, after the transition from traditional fuel vehicles to smart electric vehicles, the value of its single-vehicle chip will quickly increase from more than US$600 to nearly US$3,000.

Even if local companies actively expand production, it will take time to release the new production capacity. "Although domestic wafer manufacturers have also been actively expanding production in the past two years, considering that it takes about 3-5 years for new factories to go from planning to stable production, it is expected that the shortage of chip production capacity will still be difficult to make up in the next two to three years, or there may be a structural shortage of mature process wafer production capacity required for automotive chips." Recently, at the 4th Global New Energy and Smart Vehicle Supply Chain Innovation Conference, Wu Zhixin, member of the Party Committee and deputy general manager of China Automotive Technology Research Center Co., Ltd., said.

Wang Shengyang also believes that although the chip supply in many industries has eased significantly this year, due to the rapid increase in demand for chips in automobiles and some special requirements of automobiles for wafer processing, the supply of automotive chips may remain tight in recent years.

Not only supply cuts, but also innovation suppression

If the anti-China subsidy policy in the chip bill is intended to promote the return of semiconductor manufacturing to the United States, then the supply cut-off of EDA, high-end GPUs, etc. is intended to fundamentally curb domestic technological innovation.

As we all know, the global EDA market is dominated by the "Big Three" of Synopsys, Cadence and SimensEDA, while the IP market is dominated by the United States and the United Kingdom, which accounts for more than 90% of the market, and both are highly dependent on foreign countries. Now that the United States has announced a ban on EDA, it means that Chinese companies will face greater resistance in the future in developing high-performance computing chips.

In fact, the potential risks brought by the United States' series of "chip restrictions" to China's auto industry are far more than that. As the driving force of scientific and technological innovation, the development of the semiconductor industry is also related to technological innovation in other industries, including the electrification and autonomous driving changes being promoted by the automotive industry.

In particular, the current intelligent development of automobiles has put forward a great demand for advanced chips. Some AI chips used for intelligent driving and intelligent cockpits have adopted 7nm or even 5nm processes. "This means that we must achieve independent design and manufacturing of high-end automotive chips in the next decade or so to ensure the smooth transformation and upgrading of the domestic automotive industry. The US sanctions on high-end chips will affect the innovative development of key technologies for intelligent driving in my country. We must rely on our own efforts to achieve breakthrough development." Li Ming, general manager of JAC Group, said at an industry event a few days ago.

Taking the latest US ban on high-end GPU sales as an example, although the relevant products are currently mainly used for the training of artificial intelligence algorithms and are aimed at supercomputing centers and large data centers, it is not difficult to predict that as autonomous driving gradually enters a new data-driven development stage, major car companies and autonomous driving companies have begun to build autonomous driving training centers, and this type of chip will also have great potential in the field of autonomous driving.