Amid chip shortage, domestic automotive MCUs are fighting to break through

Editor | Zhao Jian

Since the end of 2020, the prices of MCUs from major international manufacturers have been rising.

A domestic Tier 1 manufacturer revealed to "Jia Zi Guang Nian" that the "spot price" held by Chinese agents of many international giants has exceeded 10 times or 20 times the original price. For example, the more mature MCU series products of NXP and STMicroelectronics generally cost more than 20 yuan around 2019, but by 2021 the spot price had risen to 300 to 500 yuan.

At the same time, even if you don't buy expensive spot chips, the delivery cycle of ordinary ordered chips has been frequently extended, from about 6 months in the past to an average of 12 months, and some chip companies have even given an estimated delivery cycle of 36 months. The general manager of Baolong Technology Division said that even after waiting for 12 months, the delivery situation is not ideal. "For customers like us who use a lot, the satisfaction rate is about 70%, while for small customers, the satisfaction rate may be only 20-30%."

MINIEYE and Moshi Intelligence, both Tier 1 companies in autonomous driving, have used the power of OEMs to obtain production capacity from wafer fabs. Yang Guang, co-founder of MINIEYE, told Jiazi Guangnian: "The market is in a tense situation of chip shortage. If Tier 1 companies encounter difficulties in obtaining production capacity from OEMs, then the OEMs may give more support if the OEMs intervene personally."

This kind of support is usually limited to when the supply chain is extremely tight, and there is no guarantee of immediate results. Yang Guang said, "When there are really no chips and we need to deliver them urgently, we can only buy high-priced spot goods."

The reason why MCU is difficult to buy is that both the supply and demand sides have changed due to the impact of the epidemic.

From the demand side, the development of new energy vehicles and smart cars in recent years has greatly stimulated the demand for MCU chips. As a chip responsible for "execution", a car used to need 20 to 30 chips, but a new energy vehicle needs at least 70 to 80 chips, or even more than 100 chips. For example, the windshield wipers of traditional cars are manual, but the new energy vehicles will be intelligently executed according to the sensing effect of the rain sensor, which will use more MCUs; the handlebars will automatically pop up according to the arrival of the passengers, which requires a new MCU compared to the mechanical type of traditional models; for example, the atmosphere lights inside new energy vehicles require nearly ten MCUs to control, and sometimes even more.

From the supply side, 60% to 70% of automotive-grade MCU chips are manufactured by TSMC. Due to the impact of the epidemic, the industry expects that car sales will decrease, so TSMC reduced its allocation of automotive-grade MCU production capacity in 2020. The increase in terminal sales of new energy vehicles after the epidemic directly led to a shortage of chips in the supply chain. Furthermore, since automotive-grade MCUs have extremely high requirements for safety and reliability, production line construction is more complicated than other types of chips, and generally takes about five years, and the shortage of chips is difficult to alleviate quickly.

In the final analysis, the root cause of the chip shortage is the limited production capacity on the manufacturing side, so whether it is Tier 1 or the OEM, it is doomed to be unable to guarantee supply. In addition, the reason why the automotive chip factories in the United States, Japan, and Europe have been able to develop to this day is also due to the development of local car companies. In the face of the chip shortage, they will definitely be more inclined to give priority to local car companies.

Faced with the pressure of out-of-stock, domestic automakers and Tier 1 have begun to turn their attention to the domestic market.

The downstream released the demand for MCU production capacity, and domestic chip manufacturers also hope to seize this historic opportunity. You know, the testing cost of MCU on the car is high and the cycle is long. Once certified, it will generally be supplied for more than ten years. If it were not for the shortage of chips, it would be very difficult for domestic manufacturers to enter the supply chain of car companies. Sometimes chip manufacturers have to wait in line for several years for car testing.

Many automotive chip companies that did not make MCUs in the past have launched new MCU product lines, such as Jiefa Technology, CoreChips, CHIPWAYS, Hangshun, Unigroup, BYD, CoreTitanium, Endee, and Flagchip. Among them, Jiefa, CoreChips, and CHIPWAYS all have SoC product lines, and have released newly developed MCU chips since the chip shortage.

In fact, JieFa Technology and CHIPWAYS started developing MCU before the chip shortage. Qin Ling, founder and chairman of CHIPWAYS, told "Jia Zi Guang Nian": "The design of SoC includes the control part, so it will be relatively easy for the team that designed SoC to design MCU. And for car users, compared with using only one SoC chip in a car, MCU products will be used in more scenarios." JieFa Technology believes that with the development of Chinese automobiles and smart cars, the huge automotive MCU market needs Chinese chips.

Some MCU companies that once focused on consumer and industrial MCUs have also tried to enter the automotive MCU market. Among them, listed companies such as GigaDevice and Fudan Micro have already obtained some opportunities to enter the automotive market, and startups such as Xinwang Micro, Lingdong Microelectronics, Pengpai Microelectronics, and Xihua Technology have also attracted the attention of investors in this wave after turning to automotive MCUs and obtained multiple rounds of financing.

Companies that have been focusing on automotive MCUs have even adjusted their product strategies in order to seize the opportunity to enter the automotive market. Geng Xiaoxiang, CEO of Yuntu Semiconductor, told Jiazi Guangnian, "We originally planned to enter the mid-to-high-end market, but due to the chip shortage at the time, we adjusted our plan in mid-2020 and first entered the mid-to-low-end mature market and designed the YTM32B1L product series." For Yuntu, this series quickly opened up the market and laid the foundation for the subsequent promotion of high-end MCUs.

However, getting the opportunity for on-vehicle testing is just the beginning. The road to domestic substitution of MCUs requires the cooperation of the entire industry chain.

Automotive-grade chips need to go through a complex verification process from design completion to mass production.

Specifically, from chip design to mass production, it is necessary to first produce tape-out, do chip packaging testing, and then go through DV (Design Validation) and PV (Product Verification) testing. DV mainly tests the rationality of the product design, and PV mainly verifies whether the product can meet the performance requirements. Generally, after the chip factory completes the design, the chip factory will first do the DV test. After the debugging test passes, it will be handed over to Tier1 for adaptation development and DV testing. After passing, it will be handed over to the host factory for PV testing.

For automotive chips, in the PV verification phase, in addition to testing product performance, it is also necessary to test whether the product can meet the requirements in terms of safety and reliability, such as how it reacts to extreme temperatures and how to deal with failures. These are generally referred to as functional safety tests. In addition, some OEMs and Tier 1s will also put forward requirements for chip manufacturing defect rates and product performance consistency.

In the past, the automotive chip verification phase took about 12 to 18 months. However, in the chip shortage, "time" has become the most precious resource, so the entire testing process needs to be compressed, sometimes even only 4 to 6 months.

This certainly requires chip factories to have good chip design capabilities, but more importantly, the cooperation between Tier1 and OEMs. For example, the most notable thing is that the chip verification work that originally needed to be carried out in stages has become parallel under the chip shortage. Geng Xiaoxiang told "Jia Zi Guang Nian", "When our products provide engineering samples, strategic customers have already started their designs. After our mass production samples are provided, they will do DV experiments, and after the DV experiments pass, they will cooperate with the car manufacturers on PV. So the work that originally took a year or a year and a half, we have compressed it into half a year."

In this process, Tier 1 bears certain risks. Tier 1 needs to develop an adaptation system based on the MCU. For a company that has not cooperated with them before, the development cost will be at least one million yuan, and the cost of manpower and equipment will also need to be added for several months.

"If the chip is unqualified, then all these costs will be wasted. But in the chip shortage, Tier 1 needs to weigh the risk of chip shortage or the risk of investing millions of yuan," Geng Xiaoxiang explained.

Tier 1 manufacturers have also begun to pay attention to the R&D progress of chip companies and intervene in testing earlier. Wang Xufang, Chief Marketing Officer of Unigroup, told "Jia Zi Guang Nian" that some important customers will start evaluation and verification based on FPGA chip prototypes in advance. "Alpha customers can carry out some software deployment development and test verification based on our early chip prototypes. At the same time, they will also be the first to get the chips after tape-out to start subsequent development, which can greatly shorten the development cycle of related basic software and application software."

In order to "get on board" faster, even the attitude of the OEMs has changed. Chip factories were originally only Tier 2~Tier 3 in the car company's supply chain, and they needed to go through the corresponding Tier 1 companies to supply car companies. But under the chip shortage, OEMs have also taken the initiative to communicate directly with chip factories. Dr. Qin Ling told "Jia Zi Guang Nian" that even some local governments will come to them to ask for production capacity. "In places such as Shanghai, Shaanxi, and Liuzhou, Guangxi, the automotive industry chain accounts for a large proportion of GDP, and the pressure on the government is no less than that of the OEM."

Even in some areas with low safety requirements, such as ambient lighting and windshield wipers, some OEMs have even purchased industrial and consumer-grade MCUs that have not been safety-verified. There are even consumer-grade MCU companies that did not know their chips were already in use in cars when their MCUs had problems and were recalled.

However, the core of the MCU chip shortage lies in the shortage of production capacity on the upstream manufacturing end , and various links in the industrial chain have begun to use their own methods to demand production capacity from wafer fabs.

Li Qinglu, product and marketing director of JieFa Technology, said that the company started the automotive chip business early and has maintained a very good cooperative relationship with many domestic and foreign foundries such as TSMC.

Wang Fuyu, managing director of Holly Capital, told "Jia Zi Guang Nian" that the company's partner team has decades of experience working in world-class wafer foundries. After investing in Semiconductor, Holly Capital assisted the company in establishing a good cooperative relationship with the foundry and obtaining sufficient production capacity support.

Many companies have also turned their attention to China. The OEMs in Shanghai will also help chip companies to ask for production capacity from wafer fabs in Shanghai. The CEO of a chip company revealed to Jiazi Guangnian, "For example, SAIC will help us talk to Huahong. Both automobiles and integrated circuits contribute greatly to Shanghai's GDP, and their goals are relatively consistent."

However, there is a severe shortage of automotive production lines in China, which requires chip companies and wafer fabs to develop together. Xu Chao, vice president of CoreDrive Technology, shared with Jiazi Guangnian that "the processes currently required for MCUs are mostly mature processes (40-90nm) , but due to corporate planning and policy guidance, wafer fabs have been developing mature processes below 28nm in recent years, so production capacity is in short supply." Therefore, some wafer fabs willing to produce MCUs have tried to transform the original consumer-grade and industrial-grade MCU production lines with excess capacity into automotive-grade production lines.

Geng Xiaoxiang told Jiazi Guangnian, "We worked with domestic FABs on our first mass-produced product, and modified and verified it according to automotive standards. For example, the previous industrial standards required the chip temperature to be between 0 and 85 degrees, but now the automotive standards are -40 to 125 degrees, so the production line needs to be adjusted accordingly. In addition, the IP library, EDA development tools, and chip development process of the wafer factory also need to be adjusted accordingly."

CHIPWAYS has also reached a cooperation with Tongfu Microelectronics, a domestic packaging and testing company. Dr. Qin Ling believes that this is a connection that can only be established between domestic companies. "We need to share product documents, test data, and even lower-level databases with each other."

The total mobilization of the industry chain has helped domestic MCUs gain opportunities to get on board, but compared with international giants, there is still a long way to go.

In fact, if we only look at the functional indicators, domestic MCU chips are comparable to similar international competitors.

According to MINIEYE co-founder Yang Guang, in 2021, when the chip shortage was the most severe, models such as STMicroelectronics' STM 32 series, SPC563 series, and NXP's FS32K series were all in short supply on the market. It was at that time that Jiefa Technology's products "came in as a substitute."

The substitutes come from two product series of Jiefa Technology: the higher performance AC781x series, which uses the ARM® Cortex-M3 core, and the slightly lower performance AC7801x series, which uses the ARM® Cortex-M0+ core. "Jia Zi Guang Nian" found through comparison that the parameter indicators of the two product lines are not inferior to the same product series of Infineon and NXP in terms of operating frequency, memory, operating temperature, voltage, and power consumption.

Yang Guang told Jiazi Guangnian that in terms of function selection and usage, there is not much difference between domestic MCUs and international products. The general manager of Baolong Technology Division also revealed: "Now we pay more attention to whether the functions of domestic MCUs are basically close to those of imported MCUs. We do not pursue better performance. The most important thing is to see whether the product can realize the function, and at the same time, we need stable supply and quality."

The real difference is hidden in places that are invisible at first glance . Wang Fuyu, managing director of Heli Capital, told "Jia Zi Guang Nian" that for a high-end automotive MCU chip, functional design may only take up less than 50% of the time, and most of the remaining energy should be used in "functional safety and reliability design", which is a test of skill.

As we all know, automotive chips are notoriously difficult to make. In terms of operating temperature, operating stability, anti-interference performance, service life, low failure rate and other aspects, automotive chips "completely beat" consumer and industrial control chips. Taking the fault tolerance rate as an example, the number of errors allowed for consumer chips per million chips is 300-500, but the fault tolerance rate of automotive chips is infinitely close to 0 because it involves personal safety.

A vivid metaphor is that if the work of consumer MCUs is to maintain urban order under normal circumstances, then high-end automotive-grade MCU chips must maintain the operation of cities under severe circumstances such as the outbreak of an epidemic. "When the epidemic spreads over a large area, it will greatly increase the risks of shortages of materials and personnel, which is the real test of our level," said Wang Fuyu.

The power is reflected in several details, the first of which is "cost" . Automotive MCUs need to achieve a high level of "test fault coverage", that is, to ensure that the test vectors can detect as many faults as possible to ensure that the test error rate is as small as possible.

From 99.9% to 99.99%, 99.999%... Every time the test fault coverage increases by a decimal point, the design cost will double, thus widening the gap in strength between different manufacturers. "It's like solving the same problem, others can solve it in three steps, but you still haven't solved it in eight steps. This is the difference in design methodology, which requires a long time to explore," Wang Fuyu told "Jia Zi Guang Nian".

Another major difference comes from the accumulation of "corner cases" . Corners are rare extreme situations that are often gradually exposed after large-scale deployment. Similar to the field of autonomous driving, in chip design, the "know-how" of corner cases is built on a sufficiently large amount of data.

Major manufacturers such as NXP and Infineon have generally accumulated more than 10 years of mass production experience in automotive-grade MCUs, while domestic automotive-grade MCU products first appeared after 2015.

Yang Guang said that in terms of anti-interference capabilities and other aspects, domestic chips are not as stable as international chips, and Tier 1 often needs to make more supplementary considerations when designing the entire machine, such as adding an extra layer of protection to achieve the same electromagnetic shielding effect as international chips.

In addition, due to the late start, the tool chain and ecosystem for MCU development in China are not yet mature. Currently, the software tools that support chip development are mostly monopolized by foreign platforms such as Keil and Eclipse, which brings additional adaptation costs to domestic chip manufacturers.

Wang Xufang, Chief Marketing Officer of Unigroup Core Energy, told Jia Zi Guang Nian, "Foreign giants such as Infineon have their own and third-party development tool chain support, which can achieve good compatibility and adaptation. However, to adapt foreign development tools to our own chips, we started relatively late and need to invest more time in building the tool chain ecosystem."

However, being a latecomer is not without advantages. Domestic manufacturers entering the market at this time also have some special opportunities.

More than one manufacturer emphasized its more "considerate" service attitude.

Geng Xiaoxiang told "Jia Zi Guang Nian" that once quality problems arise with foreign chips, usually one can only turn to agents rather than the original manufacturers for resolution. Their response speed and service quality are far inferior to those of local manufacturers.

Wang Xufang also explained that in the after-sales service link, once a problem occurs, it is often traced back to the two levels of application support and chip design. Most international manufacturers have application support teams in China, while the core design team is based overseas. Therefore, once the problem needs to be traced back to the design level, the efficiency of communication and feedback is usually relatively low, and it takes a long time to solve the problem.

More important than service attitude is the customized product service provided by domestic chip manufacturers. For example, a tiny detail is that some foreign MCUs will develop special interface resources for certain specific application scenarios or customers abroad, but this is not needed in China. Therefore, local chip manufacturers will make corresponding cuts to help downstream companies save costs.

Industry insiders also revealed that there is a special "national secret window period" for domestic MCUs. In terms of vehicle information security, my country has issued national secret standards. During a certain window period, it will be first opened to domestic manufacturers, and then further opened to foreign manufacturers after the standards mature. This gives domestic manufacturers a "first-mover advantage."

However, in the final analysis, whether it is faster response speed or more customized services, compared with the more fundamental high security and high reliability of chips, they are only small opportunities. To be able to become the first or second supplier in the long term, the "hard power" of product functional safety and stability still needs to be competed.

In the field of automotive chips, time is often measured in decades.

For specific functional points of the same model, the chip supply guarantee agreement is signed for at least 10 years. This is based on considerations of supply chain security and stability as well as cost factors.

If the downstream wants to replace the original imported chips from Infineon, NXP, etc. with domestic chips, it will definitely be a "whole-body" job.

Since MCUs must eventually run code tests on software, switching suppliers is not just a matter of simply transplanting chips and re-soldering them for testing, but rather redeveloping the entire design platform to adapt to new models. "Considering human and financial resources, it is normal to invest tens of millions of yuan, and the development platform cannot be reused for different original manufacturers," Wang Xufang revealed.

The closer to the functional points with high safety level such as body control and brake system, the more obvious this phenomenon of "locking in" suppliers is. According to Wang Xuehai of Moshi Intelligence, Tier 1 chip suppliers generally choose two, one as the main and one as a backup. Once the two supply points A and B are locked, it will be difficult for the following suppliers to enter.

For high-end MCU manufacturers, the situation has become clearer: amid the chip shortage, downstream companies are constrained by production capacity and are eager to change the situation where both AB supply points are occupied by international chip manufacturers, and have begun to "shift gears" intensively.

However, this window of opportunity will not last long.

The general manager of Baolong Technology Division believes that the opportunity will last for at most one year: "If manufacturers fail to bring proven stable and reliable products to market within this year, the market will be occupied by the first comers, and it will be very difficult to make efforts later."

Xu Chao, vice president of CoreDrive Technology, explained the opportunity of the "window period" from another perspective. Xu Chao revealed that as cars become more intelligent, car companies are adopting a more intelligent new platform architecture - dividing the entire vehicle into 5-6 functional areas, setting up a separate "regional controller" for each area, and using high-performance MCUs to support the core control and computing of the entire area, thereby improving the intelligence level of the terminal.

The new design architecture places higher demands on MCU performance. However, even looking around the world, there are only a handful of MCU products that can meet both high performance and high security levels. Based on this, Xu Chao judged that "within 5-10 years, based on the needs of new architectures of car companies, a batch of new high-performance MCU products will be launched on the market. This track will also produce top players and form a stable competitive landscape."

The high threshold and high investment mean that the high-end MCU market will only be a competition among a few leading manufacturers. In the low-end MCU market, the competition is much more lively.

An industry insider revealed that "all those who make industrial control MCUs, regardless of whether they have real money invested, have announced that they are entering the automotive MCU market. This can at least 'take advantage of the trend'."

Radio, ambient light, window lift, chassis control, brake control... If we consider these safety performance requirements from low to high as segments, then there are at least dozens of "segments" in a car, each with its own "know-how". Wang Fuyu described it as, "If there are 100 companies in the low-end market to share a cake, no one will get too much."

Dr. Qin Ling also believes that the low-end market will eventually present a "hundred schools of thought contending" pattern. "Because the various applications in the car are completely different, it is not like the consumer and communication categories, where one general-purpose chip conquers the world. Due to different customer needs, there will not be highly standardized products, which will eventually lead to a "hundred schools of thought contending" pattern in the industry."

The greater concern is not whether domestic players can seize the opportunity, but the chaos in the competition order caused by the influx of too many players.

Industry insiders revealed that many downstream customers have begun to try to negotiate with upstream chip manufacturers at the lowest market price, trying to lower prices. "It's hard to say whether it's a customer problem or an industry problem, but price competition is bound to happen. There's nothing you can do about it."

Some people also revealed that many domestic manufacturers are eager to use various certifications and tests to promote themselves, which may cause information interference to market judgment. "Many tests are just process tests, not real product tests. It's like a chef draws a PPT to show all the cooking processes. As for whether it tastes good or not, it still depends on the market evaluation given by customers after tasting it."

But if we take a longer-term view, the chip shortage will only be a ripple in the semiconductor cycle. Whether it is the fierce competition among mid- and low-end MCUs or the PK between high-end MCU oligarchs, market competition will eventually converge on product quality itself.

For automakers, the degree of tolerance for domestic suppliers will be a trade-off of a series of comprehensive factors: quality is definitely the first priority, and safe and stable supply capabilities are principles that cannot be compromised in the field of automotive chips. Once this bottom line is met, whether they are willing to grow with domestic suppliers will also depend on the iteration speed of domestic chips and the cost of switching suppliers downstream.

For domestic MCUs, getting a ticket is just the beginning. What is coming will be a new competition and fight.

- END -