From the end of 2020, the lack of cores has become the biggest challenge in the automotive industry. Since January last year, few OEMs have been forced to stop production or cut production. Chip shortage not only led to a decrease in the output of the automotive industry, consumer car purchase costs increased, but also affected the transportation industry that relies on freight trucks, which in turn affected global logistics costs, and even interfered with many technology companies that developed autonomous passenger cars or trucks. Therefore, understanding the ins and outs of this shortage of automotive chips, when the core shortage crisis will ease and how to solve it is a compulsory lesson for industry insiders and those involved in the semiconductor and automotive industries.
The reason for the lack of cores
The pre-process process used in automotive chips is still a mature process of more than 40 nanometers and a traditional process, which account for 95% of the fuel vehicle chips. These chip products mainly include various microprocessors, power supply chips, data link chips and interface chips used in automotive distributed electronic architectures. The emergence of intelligent electric vehicles has greatly increased the use of chips in advanced processes and cutting-edge processes, but considering the current market share of intelligent electric vehicles, the demand for chips in the automobile industry is still dominated by mature processes and traditional processes.
Demand for chip processes in the automotive industry (Source: Roland Berger)
Production side
On the production side, the expansion of production capacity in the semiconductor industry is relatively cautious, with an annual growth rate of only about 6%. Fab capital expenditure is concentrated in advanced and cutting-edge processes below 40 nanometers, and is a 12-inch production line, with a compound annual growth rate of 26% in 2020-2022. For the mature process and traditional process wafer production line that automotive chips rely on, the expansion of semiconductor suppliers is more conservative, for example, the capacity growth rate of traditional processes is only 2%.
Reinhard Ploss, CEO of Infineon Technologies, explained that the automotive industry is accustomed to purchasing chips at low prices, resulting in (semiconductor suppliers) not having the motivation to expand production capacity. Hassan El-khory, CEO of ON Semiconductor, has a similar statement, most of the automakers adopt the Just-in-Time management model, requiring chip suppliers to passively cooperate - semiconductor manufacturers spend several years, hundreds of millions to billions of dollars to establish chip production capacity, may be cancelled by the oem 30 days before production. Therefore, semiconductor manufacturers are cautious about expanding production capacity.
In my experience, for example, management requires not only an ROI of roI for expansion within 18 months, but also a confidence level of more than 80% for product line managers to forecast customer demand, otherwise they would rather give up the possible benefits from additional customer demand. Even if you agree to invest, review at least semi-annually or quarterly whether customer demand is still there. If circumstances change, contingency planning needs to be implemented immediately, including but not limited to finding replacement customers, transferring capacity to other products or product lines, canceling expansion plans, etc.
Changes in total capacity in the semiconductor industry and expansions over the years for mature and traditional processes (Source: Roland Berger)
Secondly, most of the production lines for the production of automotive chips are 8-inch lines (or even 6-inch lines), the production line is established earlier, depreciation is basically completed, so the wafer production cost is lower, and then invested in the construction of a new 8-inch fab has no cost advantage, so the production capacity of 8-inch wafers in 5 years has an average annual growth rate of only 3%. Although some semiconductor IDMs transfer the production of automotive chips to the old 12-inch production line in the intention of increasing production capacity and obtaining scale effects, it takes a long time for production line debugging, product verification (semiconductor suppliers and automotive customers) and capacity ramp-up.
Third, some existing 8-inch lines and 6-inch lines are also transforming the production of compound semiconductors such as silicon carbide or gallium nitride, further reducing the production capacity of silicon chips, which also directly or indirectly affects the production of automotive chips.
TSMC Fab2 switched from a traditional silicon process to a 6-inch silicon-based gallium nitride chip in 2015 and expanded to an 8-inch foundry line in 2021 (Source: Internet)
Finally, black swan events such as the Texas cold wave, the closure of packaging companies in Malaysia due to the epidemic, and the Renesas factory fire have further hit the semiconductor industry with extremely long supply chains, making the automotive industry, which requires the normal supply of hundreds of chips to produce on schedule, have to endure the shackles of the shortest board in the chip supply chain.
Consumer side
On the consumer side, semiconductor demand has soared since 2020, with a growth rate of 17%, far exceeding the expansion rate of production capacity. The main reason is that working from home due to the pandemic has greatly promoted the demand for personal computers and cloud computing/storage. At the same time, in 2020, OEMs canceled a large number of chip orders, allowing wafer foundries and semiconductor IDMs to transfer production capacity, especially those of automotive microprocessor chips using advanced processes, to consumer electronics and data centers.
But the recovery of the auto market in the second half of 2020 far exceeds the expectations of the auto industry, and OEMs are trying to restore previously planned vehicle production and increase investment in electric vehicles with larger chip consumption. As a result, demand for automotive chips has staged a V-shaped recovery in 2020 and 2021, even exceeding pre-pandemic levels. However, the previously allocated production capacity has not recovered, which has led to the inventory of automotive chips on all sides of the semiconductor supply chain being much lower than normal. This has led to over-ordering in the chip procurement department (Double Ordering) and even a shortage game between customers, and the resulting bull whip effect has made each level of the semiconductor supply chain more or less amplify the demand demand, resulting in a rapid increase in demand in the short term.
Monthly sales of automotive chips (Source: SIA)
In addition, automotive chips have low purchase prices and harsh procurement conditions due to the strong bargaining power of automakers, and the same production capacity can be higher profits if supplied to other customers. As a result, semiconductor manufacturers and foundries have no incentive to actively produce and supply automotive chips until multinational government departments step in. These reasons have combined to lead to this historic chip shortage.
When the lack of cores is alleviated
As to when this chip shortage will ease, Intel CEO Patch Gelsinger believes that the chip supply and demand balance may not be achieved until 2023, and Infineon Plos also believes that the crisis will continue in 2022. At the same time, some industry observers said that the chip market will achieve a soft landing in 2023 and return to the "Goldilocks" zone. Some investors in the chip industry also believe that the growth rate of chip production capacity and the growth rate of chip sales will intersect in the third quarter of this year, and the lack of cores will be alleviated. In any case, no one has a crystal ball that can accurately predict the future, but some data and indicators can help us analyze the semiconductor supply and demand situation and help determine when the chip crisis will ease.
Capacity utilization
The capacity utilization rate of semiconductor IDM and wafer foundries in the front process is an important production-side indicator, indicating the ratio of the total number of wafers produced on the production line to the installed capacity of the production line in a certain period of time. Semiconductor suppliers typically provide this data in the quarterly report or subsequent earnings call, which can reflect the supply of chips over the next 3-4 months, depending on the general production cycle of the chip product.
Generally speaking, if the capacity utilization rate is more than 80%, the wafer manufacturing production line is more fully utilized. However, since the epidemic, this data has been more than 90%, some fabs are close to 100%, and even individual production lines have reached more than 100% through high-intensity overtime production. However, if it is too rigid, it will be discounted, and the capacity utilization rate will return to normal levels sooner or later. Therefore, this indicator can judge the supply of chips in the short term.
Wafer line capacity utilization since 2019 (Source: SIA)
Chip inventory levels
Chip inventory levels are another figure that will be mentioned in the semiconductor company's quarterly earnings reports, which can generally be understood through the Days Sales of Inventory (DSI) inventory turnover data. The semiconductor industry is a cyclical industry, usually trending as an industry boom, with increased revenue and decreased inventories. Revenue growth then began to slow down after peaking, but as manufacturers expanded production, inventories increased. Subsequently, the chip gradually oversupplyed, the price fell, the revenue decreased, and the industry entered a trough period. Semiconductor companies also suspend the expansion of production or eliminate backward production capacity, waiting for sales to increase faster than production speed and digest inventory. Finally, because of the increase in chip consumption, the industry turned into a boom.
As a result, revenue growth slows or flats, but inventory increases may be a sign that demand has peaked and capacity is gradually catching up. For example, in the recent quarterly report of Core Source Systems (MPS), the number of inventory turnover days increased to 134 days, and the company's management believes that the balance between supply and demand will be reached by the end of 2022.
Inventory turnover and revenue changes of five semiconductor leaders (Renesas, NXP, Infineon, ST, and Texas Instruments) with revenue growth and decrease rates on the horizontal axis and inventory turnover on the vertical axis (Source: Nikkei Chinese Network)
In addition, garnter's semiconductor supply chain inventory tracking index can also be referred to, the latest data is currently 0.90 in the third quarter of 2021, an increase from 0.88 in the previous quarter, entering a moderate shortage zone.
Chip price ASP and delivery Lead-Time
These two figures are not usually provided directly in the earnings report, but they are often mentioned in the earnings call and are also an important indicator of chip supply and demand. For example, in the recent financial report and investor conference of Nano Micro Semiconductor, the company's management mentioned that the delivery time of its Gallium Nitride chip has not changed, and it is still 12 weeks, indicating that TSMC as the foundry of Nano Micro, the production capacity of Silicon-based Gallium Nitride can still meet the needs of Nano Micro.
The Federal Reserve Bank Economic Database also provides a production price index PPI for the semiconductor and other electronic component manufacturing industries, which can be used to understand chip price trends. In addition, some agents can also provide aggregate information on delivery dates, such as Avnet's Abacus Delivery Guide.
Production Price Index for Semiconductor and Other Electronic Components Manufacturing (Source: FRED)
Expansion of production lines
Semiconductor pre-process is one of the main bottlenecks in the current chip shortage, so any significant wafer fabrication expansion will affect chip supply in the future. Depending on the size of the project and the amount of investment, the expansion of the previous process includes the construction of a new wafer production line, the expansion of the existing production line by purchasing equipment, or the increase of manpower to improve the capacity utilization rate of the existing production line. In the current situation of high capacity utilization, the first two have become the main means of expansion.
Increased utilization, capacity increased capacity, and time required to build new wafer lines on existing wafer lines (Source: Bain Capital)
However, whether it is the construction of a new wafer production line or the purchase of equipment to expand the existing production line, it will involve fixed asset investment, and internal approval is strict. Even if approved, the current delivery time of semiconductor equipment is more than half a year, and the time spent on installation, commissioning and trial production is longer. The average time required to build a new wafer line or purchase equipment to expand capacity is 2-3 years and 1-1.5 years.
For example, the performance of a certain IGBT product that the author was responsible for was quite good after its launch, and the demand from end customers exceeded expectations, but the production capacity was invested according to the forecasts of previous customers. After the analysis of the manufacturing process, a certain process is a short capacity, as long as you buy a wafer bonding machine, you can increase the weekly wafer output WPW by 50%. However, the delivery time of this equipment is more than 9-10 months, coupled with the internal audit, appropriation, contract formulation, order placement, and the engineering service time required in the later stage, the production capacity can only be maintained within at least 15 months. Therefore, compared with the previous indicators, the expansion news of major semiconductor leaders and major wafer foundries can only provide medium- and long-term supply-side data.
However, product/process change notices (PCNs) published by semiconductor companies can provide information on short- and medium-term capacity. For example, in a product/process change notice issued by Infineon in June 2020, it was mentioned that the OptimoS 5 series will start shipping from the 12- fab in Dresden in three months, which means that the production capacity of the product has the potential to increase significantly in the short term.
Major product changes mentioned in Infineon's 2020 Notice of Product/Process Change for OptiMOS 5 (Source: Infineon)
Terminal requirements
Automobile sales reflect the terminal demand for automotive chips, if this data still maintains a growth trend and the growth rate has not slowed down, it means that the demand for automotive chips will be high in a short period of time. This can also be seen through the ratio of automobile inventory to sales in the Federal Reserve Bank's economic database. The historical average for this indicator is between 2-3, but the most recent Data for November 2021 is only 0.24.
Ratio of inventory to sales in the U.S. auto market (Source: FRED)
In addition, you can refer to the Manheim Used Vehicle Value Index, which is updated every six months, which is also an indicator commonly used by ARK Investment to analyze the demand for automotive chips.
Mannheim Used Car Value Index (Source: Cox Automotive)
How to alleviate
The beginning of November last year was the deadline for the U.S. Department of Commerce to ask semiconductor companies for internal information on their product types, processes and capacities, inventories, supply chains, and more. According to information released by the Bureau of Industry and Security under the U.S. Department of Commerce, hundreds of industry chain-related companies have submitted the required information, including qualcomm, Apple, TSMC and Cisco, and most of the industry's important players. Although information involving corporate trade secrets is not displayed to the public, some companies give their own views on the information revealed.
The U.S. Department of Commerce requires semiconductor-related companies to provide information, taking information disclosed by NXP as an example (Source: US BIS)