1. Interpretation of Import Tariff Policy
Changes in certification standards: Following the latest customs tariff changes, the certification of chip origin has shifted from being primarily based on the production location of the complete chip package to being certified according to the wafer fabrication location. This change is strictly enforced, leaving little room for ambiguous operations, as chips have clear traceability information, making it difficult to manipulate during customs clearance.
Potential exemption possibilities: The pace of policy changes is rapid. In the context of reciprocal taxation, the U.S. has exempted certain products, and China may reciprocate by exempting chip products, which would positively impact the actual implementation of tariffs. However, it is necessary to monitor for new policy changes in the next 1-2 weeks. Currently, the customs policy categorizes by tariff number, with integrated circuits certified based on wafer fabrication location. In cases where the value added exceeds 30% during packaging and testing, this more often points to non-integrated circuit products, such as diodes and power devices.
2. Automotive Chip Market Analysis
Value and distribution of chips per vehicle: Taking a 400V electrical architecture new energy vehicle as an example, the total value of electronic chips in the vehicle is approximately $2000. By application area, the cockpit and autonomous driving domains have higher chip values due to rich functionalities and complex units, each accounting for over 25% and nearly 30%; the chassis and power domain, along with the three electric systems, account for about 30%; the body domain accounts for about 15%.
Domestic substitution situation: Most general power products have domestic alternatives; simple operational amplifiers and communication protocol chips in the signal chain are relatively easy to substitute domestically. However, high-end complex digital processing chips (such as high-end MCUs and large SoCs), high-speed signal chain analog chips (such as SerDes chips), and some products limited by processes that cannot achieve high yield mass production (such as high-side drivers and certain motor drivers) are still dominated by foreign solutions. Overall, about 70% of analog chips have domestic alternatives, while about 50% of digital chips are dominated by foreign investment.
3. Impact of Tariffs on Vehicle Manufacturing
Feasibility and timeline for domestic replacement: If tariff policies are implemented, taking TI and other American analog chip manufacturers as examples, about 70% of analog chips can be replaced domestically. These domestic chips already have mature mass production products and have been validated or mass-produced for some end customers, with competitive pricing. The replacement timeline is relatively quick, requiring 6-9 months, depending on the attitude of suppliers and OEMs towards change validation. For the remaining 20%-30% of products, some domestic solutions exist, but there are few mature mass production cases, and the switching risk is high, requiring policy support. About one-third of products have significant domestic gaps, requiring 2-3 years to catch up.
Current status of localization in automotive companies: Leading manufacturers have an average localization rate of about 20%. State-owned automotive companies (such as FAW, Changan, etc.) have localization requirements, with specific models achieving localization rates of 70%-80%, and overall vehicle localization reaching nearly 30%; private enterprises (such as BYD, Chery, Geely) consider supply chain security and have localization backups, with localization rates close to the industry average of about 20%; new force automotive companies (except for Leap Motor) are not keen on localization and rely on foreign chips, with localization rates just above 10%. Leap Motor, driven by cost considerations, is expanding localization relatively quickly; joint venture automotive companies have no localization rate requirements except for some recommendations within their systems, with actual localization penetration below the industry average, around just above 10%.
4. Impact of Leading Chip Suppliers
Impact on fabless manufacturers: Fabless manufacturers like NV and Qualcomm, due to supply chain cost considerations, have most of their production outside the U.S. For example, Qualcomm and MediaTek have some production in the U.S., but more in Asia through TSMC. American analog chip fabless manufacturers like MPS have supply chains in China, so the policy’s impact on them is limited and will not face direct shocks.
Impact and response of TI (IDM): As a deep IDM, TI’s analog chips account for about 80%, with over 70% of wafers fabricated in U.S. factories, theoretically making them significantly affected, involving about 60% of total revenue. Their response strategies include waiting for policy implementation and transferring some feasible production capacity to factories outside the U.S., such as the 8-inch factory in Germany, the new integrated 8-inch factory in Chengdu, and two factories in Japan. However, due to factory size and process limitations, the feasible transfer is about 50%. Over 80% of TI’s digital chips are produced in South Korea and by TSMC, with minimal impact from tariffs.
5. Automotive Companies’ Responses and Trends in Domestic Substitution
Experience of automotive companies in response: After experiencing a chip shortage, automotive companies are now better prepared for the current situation, including the implementation of mature domestic substitution plans, strengthening channel control, regular communication with chip manufacturers, and stocking through their own channels, allowing them to respond temporarily using their own means.
Expectations for domestic substitution:Driven by policy, the domestic substitution of mature general-purpose devices such as general power chips, low-speed signal chains, and general logic chips is expected to accelerate, potentially approaching 50%. In specific fields such as high-speed signal chains, BMS analog front ends, and dedicated MCUs, although domestic products have gaps, automotive companies will accelerate the introduction of domestic products for validation due to geopolitical considerations. The domestic substitution of cockpit chips is limited in the short term due to Qualcomm’s advantages, while there is some room for domestic substitution of autonomous driving chips, which is slightly better than that of cockpit chips. Automotive companies will cultivate more domestic suppliers to increase the substitution rate.
6. Other Related Issues
Value and substitution of TI analog chips: If a 200,000 yuan pure electric vehicle uses all TI analog chips, the value is approximately $600. Non-American foreign manufacturers can theoretically replace about 90% of these chips, but validation for vehicle integration will take 6-9 months, and changes to the controller layout will extend the timeline. TI’s analog chips are widely used in power, steering, dynamics, and body control scenarios.
Wafer fabrication locations and tariff impacts: Most American analog chip suppliers like TI, following the IDM model, have most of their wafers fabricated in the U.S. Some fabless manufacturers have dispersed fabrication locations, such as ADI in both the U.S. and South Korea, while MPS primarily operates in Asia and collaborates closely with domestic Huahong. Digital chips from Qualcomm and NVIDIA are mainly produced by TSMC and Samsung. Only chips fabricated in the U.S. are affected by tariffs.
Classification and value of analog chip substitution: Among analog chips, separated power supplies (such as LDOs, DC-DC converters, etc.) and general signal communication protocol chips (such as CAN, LIN, etc.) are easier to substitute, with the per-vehicle value of power supplies estimated at about $100-$150, and general signal chains at about $70-$80. Integrated power supplies and high-speed signal chain processing chips are more challenging to substitute, while driver chips are in a transitional phase for domestic substitution, accounting for more than half of the total value of analog chips.
Impact of specific chips and self-research thresholds: TI’s DMD (DLP) chips used for HUDs are currently monopolized by TI, likely produced by TSMC, although there are plans to shift to a factory in Utah, the current production is mainly outsourced. Since it is a 12-inch process and technically sensitive, there are certain risks. However, HUD is not a necessary function and can be implemented in other ways. The gap between domestic and foreign IGBT power devices is relatively small, with leading suppliers mainly in Europe, and many have domestic foundry collaborations, resulting in a smaller impact from tariffs compared to analog chips. Self-research in analog chips requires significant investment, and automotive manufacturers need to reach a scale of nearly one million units for it to be practically effective, while self-research in high-end digital SoC chips generally requires a scale of 500,000 to 1,000,000 units.
Q&A
Q: What is the current certification status of imported chips, and is there a potential for exemptions?
A: Previously, the definition of chips or integrated circuits had ambiguous space, and the common practice was to consider the production location of the complete chip package, i.e., the location of the packaging and testing factory, as the country of origin. However, following the latest customs tariff changes, chips must be certified based on the wafer fabrication location, and strict enforcement leaves little room for manipulation, as chips have complete and clear traceability information. However, the pace of policy changes is rapid; yesterday, I saw that the U.S. exempted some products from reciprocal taxation. According to the principle of reciprocity, China may also exempt tariffs on the same products imported from the U.S., which could positively impact the actual implementation of tariffs, but we need to observe whether there are new policy changes in the next 1-2 weeks.
Q: If the value added during packaging and testing occurs mostly outside the U.S., can these chips be imported smoothly without tariffs?
A: According to current customs policy, integrated circuits are certified based on the wafer fabrication location. If the value exceeds a certain percentage (e.g., 30%) during the packaging and testing phase, it points more towards products under other tariff numbers, mainly referring to non-integrated circuit items like diodes and transistors that involve more packaging and welding processes. When the added value or output from the packaging phase reaches a certain percentage, the country of origin can be redefined to other phases. However, for analog chips and mixed-signal chips, the definitions are currently quite clear, but policy changes are uncertain.
Q: What is the core chip value and domestic substitution degree used in different automotive segments (engine, ECU, three electric systems, cockpit, etc.)?
A: For a typical new energy vehicle with a 400V electrical architecture, the total value of electronic chips (both digital and analog) in the vehicle is approximately $2000. By major application areas, the value of chips in the autonomous driving and cockpit domains is relatively high, close to or exceeding 25%, nearly 30%; the chassis and power domains, along with the three electric systems, account for about 30%; the body domain accounts for about 15%. In terms of domestic substitution, general power products have domestic alternatives; simple operational amplifiers, comparators, and communication protocol chips also have domestic substitutes. Currently, foreign solutions still dominate in three main areas: high-end complex digital processing MCUs and large SoCs with high safety requirements; high-speed signal chains and specific application analog chips; and products that cannot achieve high yield mass production due to process limitations, such as high-side drivers and certain motor drivers. In terms of value, about 70% of analog chips have domestic alternatives, while 30% have limited feasibility for domestic substitution in the short term; about 50% of digital chips are strongly dominated by foreign investment.
Q: If the import tariff increase policy is implemented, what is the additional value added to the vehicle manufacturing end, how long can domestic switching be achieved, and which parts cannot be switched?
A: Considering only analog chips, about 70% can be replaced domestically. These domestic alternative chips already have mature mass production products and have been validated or have a certain market share among some automotive end customers, with cost competitiveness aligning with foreign manufacturers. If the policy is implemented, switching them to customer product designs would require a relatively quick pace of 6-9 months. If market demand is urgent, the time can be further shortened, depending on how suppliers and OEMs handle change validation. The remaining 20%-30% of products have solutions for about two-thirds domestically, but there are few mature mass production cases, and the risk of customer switching is high, requiring suitable opportunities or policy support. About one-third of products have no equivalent functional replacement domestically, with a generational gap compared to foreign products, requiring 2-3 years to narrow the gap from a product development perspective.
Q: Among joint ventures, independent, and new force automotive companies, which type currently has a more advanced localization of chips? What is the current localization rate of automotive analog chips?
A: Currently, leading manufacturers can achieve a localization rate of about 20%, which is relatively average. State-owned or central enterprise-backed automotive companies, such as FAW, SAIC, GAC, Changan, Dongfeng, etc., have localization requirements and will choose specific models, requiring full domestic solutions for some large controllers, achieving localization rates of 70%-80% for specific models, and overall vehicle localization rates approaching 30%. Private enterprises like BYD, Chery, and Geely consider supply chain security and have localization backups, pushing domestic suppliers to lower costs, with localization rates close to the industry average of about 20%. New force automotive companies, such as NIO, Xpeng, and Li Auto, are not particularly keen on localization and heavily rely on foreign chips, with localization rates just above 10%. Leap Motor, similar to BYD, is cost-driven and is expanding localization relatively quickly. Joint venture automotive companies generally have no inclination towards localization, except for possible recommendations within the SAIC and GAC systems, with actual localization penetration below the industry average, around just above 10%.
Q: How significantly does TI, as an IDM, get affected by import tariffs, and what alternative plans are there?
A: From the perspective of capacity distribution, about 80% of TI’s analog chips are produced, with over 70% close to 80% of wafers fabricated in U.S. factories, especially for new products using the latest 12-inch wafer production lines in the U.S. Theoretically, about 60% of TI’s total revenue is affected by the country of origin being the U.S., which is a significant impact. The current strategy is to wait for policy implementation, as policy changes are rapid; secondly, to create backups. As a deep IDM, TI is unlikely to outsource its analog business to other wafer foundries to avoid U.S. origin. It is more likely to transfer feasible production capacity to its other wafer factories, such as the 8-inch factory in Germany, the Chengdu factory (which is an 8-inch integrated new production line this year), and two factories in Japan. However, these factories are all 8-inch, not 12-inch, and some cost pressures and the focus on high cost-performance products will lose cost advantages when shifting from 12-inch to 8-inch. Additionally, some complex analog products produced in the U.S. may struggle to ensure yield and mass production in 8-inch factories. Overall, about 50% of TI’s production can be transferred from the U.S. to non-U.S. factories.
Q: Where are the main fabrication locations for TI’s autonomous driving-related chips (such as TDA4) and will they be affected by import tariffs?
A: This part is relatively less affected. For example, TDA4, cockpit SoCs, and other digital products, over 80% of TI’s production is currently outsourced. TDA4 is mainly fabricated in South Korean wafer factories, with some production at TSMC; DR821TBS is mainly produced in South Korea and Taiwan. The Micron factory in Utah acquired by TI has not fully switched processes yet, which may have some impact, but overall, over 70%-80% of TI’s digital products are produced outside the U.S., so the impact from tariffs will not be particularly significant.
Q: How does the mindset and content of automotive companies facing related issues differ from the chip shortage in 2021?
A: Automotive companies currently have more experience in response and are better prepared compared to before. On one hand, some mature domestic substitution plans have been validated or implemented in mass production vehicles; on the other hand, automotive companies have strengthened channel control, and many now regularly communicate with chip manufacturers for delivery, and have dedicated distributors and suppliers for additional stocking, allowing them to respond temporarily using their own means rather than relying more on Tier 1 suppliers.
Q: From a relatively neutral perspective, which areas will definitely see an increase in domestic substitution rates this time?
A: First, mature general-purpose devices, such as general power chips, relatively simple low-speed signal chains, and general logic chips, can be produced by domestic manufacturers, with no significant product gaps compared to foreign solutions. If policies push the industry towards domestic substitution, the inclination and preference for these products will accelerate, potentially approaching 50%. Second, specific chips previously monopolized by American companies, such as high-speed signal chain SerDes products, BMS analog front-end AFE, some dedicated MCU chips, and Ethernet-related chips, will see automotive companies accelerate the introduction of domestic products for validation, even if domestic alternatives are not yet mature and have product gaps due to geopolitical considerations. For driver and isolation products in analog chips, besides TI, there are also European suppliers like Infineon, ST, and NXP, whose products are mature, have large market shares, and have domestic supply chain layouts. The urgency for customers to switch to domestic products is relatively low, and they will prioritize solving issues in areas dominated by American companies.
Q: Will the import tariffs substantively promote the share of domestic chips in the cockpit?
A: The promotion is relatively limited. Currently, the main advanced products in the cockpit are still based on Qualcomm solutions, and there are not many domestic solutions that can directly compete with Qualcomm in the short term. For example, while Chipone is performing relatively well, many Chipone projects were initiated due to domestic policies, and in pure commercial competition, Qualcomm still holds an advantage, with end vehicle manufacturers being quite reliant on Qualcomm solutions. Additionally, Qualcomm has a large business scale in consumer electronics and will leverage its capabilities to minimize the impact of tariffs, continuously focusing on the Chinese market, including automotive.
Q: If a 200,000 yuan pure electric vehicle theoretically uses all TI analog chips, what would be the approximate total value?
A: If all analog chips are from TI, the estimated value would be around $600, which is several hundred dollars.
Q: If we do not consider domestic substitution, what proportion of non-American foreign manufacturers can replace TI analog chips?
A: In the analog chip field, there is rarely a single dominant player; many products are monopolized by TI and ADI. It is estimated that about 90% of TI’s analog chip products can find alternative products from non-American companies, but substitution requires time for validation, vehicle integration, and updates.
Q: In which common scenarios are TI’s chips used in vehicles?
A: First, in power applications, across all application areas, whether for small sensors or large core dynamics, chassis, three electric systems, including autonomous driving body control processors, TI has the largest share among power suppliers; second, in specific application scenarios, such as steering, core dynamics, body control involving external sensor drives for motors and valves, full-bridge and half-bridge drives, as well as large loads (such as lights, air conditioning, heating devices, etc.) in vehicles, TI’s usage can account for a significant proportion; third, in signal products, such as isolation interfaces, isolation operational amplifiers, standard CAN, traditional operational amplifiers, high-speed signal chains, and analog front ends (such as Ethernet), TI has products that can cover these areas.
Q: From the perspective of tariff fluctuations, what is the current strategy of automotive companies regarding chip localization?
A: The strategy of automotive companies is to accelerate localization. In the past two years, part of the reason for automotive companies to promote model localization was due to political pressure, considering the risks of non-independence. Even if tariff exemptions are only a short-term change, the actual impact is not significant, but overall, automotive companies will drive themselves to pursue localization.
Q: In the Chinese automotive sector, where are the fabrication locations for analog chips, Qualcomm smart cockpit chips, and NVIDIA intelligent driving chips?
A: The situation of American analog chip suppliers varies. Most suppliers following the IDM or deep IDM model, like TI, have most of their wafers fabricated in the U.S. For fabless American analog chip manufacturers, relatively few wafers are fabricated in the U.S., such as ADI having some in the U.S. and some in South Korea; MPS is a completely fabless company, with many wafers fabricated in Asia, including domestic collaborations with Huahong, which is its largest partner. Digital chips like Qualcomm and NVIDIA, which are high-end processing SoCs, are primarily produced by TSMC, with some in factories like Samsung in South Korea.
Q: Since most automotive-grade analog chips come from American suppliers, if we start using non-American analog chip suppliers, will there be certification issues?
A: There are no certification issues. European suppliers like Infineon, NXP, ST, and most domestic suppliers have been certified or used by end customers. Previously, American suppliers had advantages in terms of cost-effectiveness and commercial considerations, with customer applications and designs being more mature. If switching is required, new solutions will need validation, but the suppliers are already within the supply system.
Q: From the perspective of the Chinese automotive supply chain, will the Chinese government’s imposition of tariffs on chips originating from the U.S. have a significant impact on the entire automotive supply chain in the short term?
A: Although the policy has ambiguous and flexible handling space, it will certainly have an impact on the industry. End customers will accelerate the validation speed of non-American, especially domestic suppliers, hoping to find suitable suppliers for key components and make preliminary reserves within a year. They will also require chip manufacturers, traders, or channel suppliers in the supply chain to accelerate the search for alternative solutions outside of American products. Additionally, government support for domestic chip policies will likely have clearer new tendencies.
Q: Which automotive chips are relatively easy to substitute?
A: Analog chips can be roughly divided into two categories: power and signal chains. In terms of power, those that are relatively easy to substitute or have many mature alternatives are concentrated in separated power supplies, such as standalone LDOs, DC-DC converters, or ordinary isolated power supplies. Domestic analog manufacturers have a foundation in consumer or industrial fields, making substitution easier. However, integrated power PMICs, used in large domain controllers for autonomous driving or cockpits, and complex power control chips with high functional safety requirements, have domestic alternatives but are not as widespread as separated power supplies. In terms of signal chains, common domestic substitutes include general signal communication protocol chips like CAN and LIN, general logic chips, or operational amplifiers priced at around $0.20 to $0.30; there are also some chips transitioning from consumer protocols to automotive applications, such as charging and discharging chips, protocol chips, and Ethernet chips, which are also relatively easy to substitute. However, those with high protocol barriers or specific to automotive applications, such as SerDes, IP, or PD link high-speed signal chain processing chips, are not as easy to substitute.
Q: How is the value of the relatively easy-to-substitute chips?
A: These chips are not particularly high in price but are used in large quantities. The estimated value of easily substitutable power supplies per vehicle is around $100 to $150; general signal chain chips, such as CAN chips and logic chips, are estimated to have a value of several tens of dollars, around $70 to $80.
Q: How is the domestic substitution situation for driver-type bridge drivers and high/low-side driver chips?
A: These types of chips have domestic alternatives but are in a transitional phase and are not particularly mature. They have certain process barriers because they involve merging two different IPs (power and control) into one chip, which has process requirements. In the total value of analog chips, this category, along with those that are not particularly convenient to substitute, may account for half or more. However, domestically, there are at least some alternatives, but they lack mature mass production case support.
Q: Will IGBTs be affected this time?
A: IGBTs are affected much less than analog chips. The gap between domestic and foreign power devices is relatively small and mature, with many power manufacturers; leading suppliers are concentrated in Europe, with factories either in local IDM or many collaborating with domestic foundries, such as ST’s joint venture with Sanan, which mainly produces power devices.
Q: What scale of self-research is required for analog and digital chips to reach the corresponding threshold?
A: In the analog chip field, self-research is relatively difficult for automotive manufacturers under current industry conditions. Taking BYD as an example, its semiconductor division is large, but it has relatively few application scenarios for self-supply in the analog field. In terms of power chips, based on the current industry observations, a scale of nearly one million units per year may be necessary for practical effectiveness. For analog chips, the investment is larger, and there are more third-party suppliers to choose from, making it impractical for automotive manufacturers to invest in self-research for most application scenarios. For digital chips, such as higher-end SoC chips, new force manufacturers invest in self-research not primarily for short-term cost savings; generally, a scale of 500,000 to 1,000,000 units is required. New force automotive manufacturers that rely heavily on large processors may consider self-research, such as NIO, Xpeng, and Li Auto, which have started early in self-research for large chips. Overall, a scale of around one million units is needed for practical significance.