W-shaped trend, recovery may be delayed by one year (SMIC 3Q23 conference call)

SMIC Releases Q3 2023 Earnings Report:

On the evening of November 9, 2023, SMIC (0981.HK/688981.SH) released its Q3 2023 earnings report (as of September 2023) after the Hong Kong stock market closed. The key points from the conference call are as follows:

Conference Call Highlights:

1. Management Report

In terms of the market, there was no V-shaped or U-shaped rebound that was expected at the beginning of this year. The overall market remained at the bottom, showing a W-shaped trend. In the Chinese market, the high inventory issue that emerged last year has been alleviated and reduced to a relatively healthy level. Terminal and OEM manufacturers have managed the supply chain and planned for localization. The demand for new products from Chinese customers is good, and there is a shortage of some bulk products. However, this has also had a negative impact on older products, leading to slower digestion of remaining inventory and intensified competition in terms of homogeneity and low prices.

The speed of inventory turnover and chip conversion varies in different regions globally. In the mobile consumer and industrial sectors, Chinese customers have basically reached a balance between inflow and outflow, but the inventory of American and European customers remains at historically high levels. In addition, the high inventory of automotive products, which has been tense for three years, has prompted major customers to make adjustments to the market and tighten their orders. After experiencing ups and downs over the past year, customers have shifted from aggressive expansion two years ago to a more focused approach this year, emphasizing core business and R&D investment, stricter control of inventory and costs, and more cautious wafer orders.

Geopolitical factors have brought repeated construction and supply chain uncertainties to the industry, resulting in the gray rhino effect. Every link in the industry chain is exploring strategies and paths. We are also continuously observing and experimenting cautiously.

Due to the industry chain restructuring caused by geopolitical factors and cost competition, our customers have increased their market share, driving the growth of our business. In Q3 2023, the company's sales revenue reached $1.62 billion, a 3.9% increase QoQ, which is in line with guidance. The gross margin was 19.8%, a decrease of 0.5 percentage points compared to the previous quarter. The overall shipment volume of the company continued to increase, with a 9.5% QoQ growth. Due to the uneven recovery of different product demands, the proportion of bulk standard products is higher, leading to a decrease in the average wafer price. This situation is consistent with the report we made to everyone in the previous quarterly performance briefing.

In terms of sales revenue by region, the destocking pace overseas, especially for mobile and consumer ICs, started later than in China and is still facing difficulties. This has further increased the proportion of revenue from the Chinese market to 84%, while other regions account for 16%. Sales revenue can be classified by service type, with wafer revenue accounting for about 90% and other revenue accounting for about 10%, maintaining a stable proportion.

In terms of application classification, smartphone, Internet of Things, consumer electronics, and other revenue account for 26%, 12%, 24%, and 38% respectively. The demand for image sensing, image signal processing, RF, and Bluetooth in domestic terminals is good, with a MoM growth of 24% and 28%. Display driver inventory has returned to a healthy level, and sales revenue continues to grow by 16%. Special storage demand is strong, especially for NAND flash, which has grown by 27% MoM. In the third quarter, 8-inch revenue has slightly rebounded to 26%, an increase of one percentage point, but the capacity utilization rate is still lower than that of 12-inch, which accounts for 74% of the revenue.

This year's capital expenditure is mainly used for expansion and production infrastructure. The company's expenditure in the third quarter was $2.1 billion, totaling about $5.1 billion for the first three quarters. Considering the increasing complexity of geopolitical influences on equipment delivery cycles, in order to ensure the production of ongoing projects, the company allows equipment suppliers to break through ahead of schedule. As the transaction cycle of the global equipment supply chain has improved, the number of equipment entering the market by the end of this year will increase significantly compared to the original forecast. The annual capital expenditure is expected to be adjusted to around $7.5 billion.

As of the end of the third quarter, the company's equivalent 8-inch monthly capacity has increased to 796,000 wafers, an increase of 82,000 wafers from the beginning of the year. Due to the larger total capacity as the denominator, although the MoM increase was observed at the beginning of the third quarter, the average utilization rate still decreased by 1.2 percentage points to 77.1%.

So far, apart from high-performance computing chips related to data centers, advanced packaging such as backside processing and dual-wafer and triple-wafer copper-copper bonding, and chiplets, there are no new driving forces and highlights in the market. At the end of the year, everyone is more cautious and prefers to focus on future growth and breakthroughs.

Guidance:

The company expects to maintain steady growth in the fourth quarter, with a slight increase in sales revenue. The gross margin will continue to be under pressure from capacity depreciation, expected to be between 16% and 18%. In recent years, the company has continued to expand and make high capital expenditures, resulting in increasing depreciation expenses and heavier pressure on the gross margin.

Looking ahead to next year, we see that the market has stabilized, and the demand for mature foundry services will increase due to inventory reduction. However, there are no significant growth drivers and highlights, and we still need to wait for the recovery of the global macro economy. We believe this is the fundamental situation for this year.

2. Q&A

Q1: Regarding the question about demand, you mentioned that the demand recovery is not a V-shape or U-shape, but a W-shape. Can you explain how to understand the W-shape? In the third quarter, the revenue of many domestic design companies has rebounded, and the gross margin has also improved. I remember Mr. Zhao mentioned the term "quick freeze and emergency stop" in last year's performance meeting. How sustainable and strong is the demand from these domestic customers? A1: The first "W" refers to the previous expectation of a one-year downturn and recovery by the end of the year, but now it seems that this situation will be doubled and last for two years.

The second "W" refers to a small market rally that the central government will initiate. As you mentioned earlier, in the third quarter, we saw some small-scale design companies related to smartphones achieve historic highs in terms of profitability and sales.

In our view, the current small market rally is driven by smartphones. Many specific groups of people have not changed their phones for several years, so this round of phone upgrades is not due to many exciting features, but mainly because these specific groups of people need to change their phones. We believe that this small market rally will not significantly impact the overall situation, as the overall sales volume of smartphones in the coming year is expected to remain relatively stable compared to before.

The sales of smartphones are seasonal, with high sales during the release months and seasons, and inventory periods. However, if we look at the past two years, the overall sales volume has remained relatively stable, without significant growth. This is our opinion.

There will be some growth next year, mainly because the rapid freezing and sudden stop of last year's economy started in June and July. At that time, Europe and the United States were still increasing orders, and there was a supply shortage. The products that could be sold this year have basically been sold. If they cannot be sold, they will not be sold quickly. Therefore, it will not have a significant impact on the overall situation.

Therefore, we believe that the overall trend is smooth. The companies that have caught up with the smartphone trend are experiencing good revenue, and other smartphone manufacturers will also start stocking up, so the trend may continue. However, this is only for these companies, and the overall industry should remain relatively stable.

Q2: Regarding capital expenditure, the company has increased its capital expenditure to 7.3 billion. According to customs data, China's imports of semiconductor equipment reached 5 billion US dollars by September, which is close to the scale of the expenditure before the company's increase. Many domestic manufacturers are importing semiconductor equipment. How do you view the supply and demand relationship in the domestic mature process market? How do you view the expansion pace next year, considering that capital expenditure has already reached such a high level this year?

A2: Our mature nodes are factories that are designed to operate for more than 20 years. Therefore, when we build these factories, it is best to have one factory per year. Some factories catch the peak period of the industry, while others catch the bottom, just like the current situation. However, regardless of the timing, the operation and gradual growth of the company in the next 20 years will be carried out in both the off-season and the peak season. Therefore, it will not have a significant impact on the overall development.

The reason we mentioned earlier is because of the recent situation at SMIC. Many of the equipment we need require permits, which means that we cannot accurately predict the arrival time of the equipment. The best solution is to apply for all the equipment together and have them delivered together. If the supplier can ease the market and deliver the goods quickly, we allow them to come in first according to their delivery time. This has resulted in an increase in the previously predicted CAPEX by the end of the year due to the arrival of more equipment than originally forecasted. In response to what you just said, it seems that there are other factories also building production capacity. Will there be an excess of production capacity and competition in the future? Currently, the geopolitical situation is a hot topic, and every region is independently expanding its own production capacity. Therefore, my personal understanding is that if the demand does not expand as quickly as the production capacity, there will be an oversupply of production capacity globally, which will take a long time to digest the capacity that has been hastily built in recent years.

However, from another perspective, there are some particularly large market areas, such as the Chinese market and the US market, where the local production capacity is not sufficient to meet the requirements for complete machines and vehicles. Therefore, when calculating and working with customers on alliances and training, the production capacity that we, SMIC, are building is based on prior communication with customers and their intentions to produce a certain volume at SMIC in the future.

Based on this kind of communication, we believe that there will be a significant need for local manufacturing of semiconductors in China in the future. These customers also have strategic intentions to cooperate with SMIC. Therefore, the production capacity that we are building at SMIC is relatively high, as we anticipate future customer demand and orders.

Q3: What is the revenue proportion of specialty platforms, and what is the future layout for specialty platforms?

A3: Approximately 30% of SMIC's revenue comes from standard logic circuits, while the remaining 70% comes from specialty platforms. According to our own classification, we can divide it into eight major submarkets, such as IoT, ultra-low power, connectivity, RF, high voltage driver DDIC, CIS, specialty memory, MCU, etc. Together, these eight categories account for 70% of our revenue. However, in the current quarter, the proportion of logic circuits will be slightly higher, at a ratio of 60:40.

In the future, we will further refine the classification of logic circuits. Currently, approximately 40% or slightly more is attributed to logic circuits, while the rest is specialty platforms that have special bindings with customers and require dedicated equipment.

This is a development that is certain to happen in the future because, as you know, logic circuits evolve rapidly with the pursuit of lower costs, better performance, and lower power consumption. It is difficult to achieve the same level of progress in standard logic circuits without better size reduction. However, this is not the case for specialty platforms.

For example, when we develop 0.18 and 0.15-micron aluminum processes, everyone is also working on BCD, power, and high-voltage technologies. Every year, we need to reduce the die size, shrink the volume of LDMOS, and minimize leakage current and Ron parasitic resistance. There is a general expectation and progress in this area. Therefore, the BCD and Analog technologies we introduce at 0.18 and 0.15 microns each year are different and show advancements. CIS has undergone significant changes. Even the leakage requirements for camera lenses vary every year. We need to develop new isolation technologies to meet these demands, right? Nowadays, there is a high demand for large pixel size and high density. This has led to the transition from the previous Wafer to Wafer bonding to double CIS wafer bonding, along with ISP bonding of three wafers, and so on. These are all advancements in special processes.

Currently, SMIC specializes in serving many small customers in special processes. We focus on deep cultivation and work closely with each customer to achieve a high level of product diversity and maturity.

With the establishment of so much production capacity, mainly in China, SMIC's future success relies on comprehensive coverage of our own technological capabilities and being at the forefront. We also prioritize strategic partnerships with customers and provide excellent service quality. Additionally, we can leverage our cost advantage to compete strongly in both domestic and international markets. By focusing on these three aspects, we are confident in our ability to manage and mature these newly established production capacities.

In the field of electric vehicles, various products are needed, and in the future, there will be integrated optoelectronics in multi-layer packaging for communication. SMIC is also exploring areas that we have not previously entered, such as high-voltage and high-power components used in cars, as well as RFSOI for RF applications. We are conducting research and development in these areas.

Q4: How do we forecast the company's long-term gross margin? Does depreciation have a significant impact in the short term?

A4: SMIC's long-term gross margin is relatively easy to assess. Approximately 75% of our CAPEX is used to purchase equipment, while the remaining portion is allocated to land acquisition, factory construction, infrastructure, and research equipment. The equipment has a depreciation period of 5 to 7 years, and after about 6 months, it enters the depreciation phase. We announce our CAPEX every quarter, and it accumulates accordingly.

The second factor is our loading and utilization rate. This quarter, our utilization rate was 77.1%. We cannot predict how much our revenue will grow in the next quarter, but we can gauge it based on the utilization rate. If we are at full capacity, we can calculate the gross margin. If we expect the utilization rate to remain around 77.1%, we can estimate that the gross margin will decrease slightly each quarter. Currently, our quarterly gross margin is 19.8%, and we anticipate it to be around 16% to 18% in the next quarter, indicating a decrease of approximately 2% or slightly more. This trend continues each quarter if the loading remains unchanged and cannot be filled to capacity.

Q5: You mentioned that the recovery of demand will be delayed by one year. Is your estimation of supply-demand balance based on observations of overseas macroeconomics or the redundancy caused by geopolitical factors and duplicated capacity? If the recovery time is delayed by another year, will it have an impact on our overall expansion progress for 2024? A5: We have two particularly strong pieces of data in our hands. The first is that we have access to the inventory of our customers, including those in the top 20 or top 30 in the market, who outsource their semiconductor production to us. By looking at their inventory and the number of days of stock they have on hand, we can roughly estimate whether they will need to replenish their stock and how much they will need to replenish. We also know the percentage of market share that Semiconductor International holds in this regard.

The second piece of data is the list of products that will be launched next year, which is already in the hands of Semiconductor International. We refer to these as "new tape outs" or NTOs. These new tape outs are new products that will be ordered and produced next year. The process of securing a design win for these products takes a considerable amount of time, at least 9 months. Typically, the customer provides us with the product, we help them create the mask, then we proceed with the silicon production. After the production is completed, the product undergoes testing and validation before being delivered to smartphone manufacturers, automobile manufacturers, and industrial manufacturers. Once the product has passed all the necessary validations, the customer will place an order. This entire cycle usually takes about a year. Therefore, by November or December of the following year, the products that will be ordered are already in Semiconductor International's factory. At this stage, we are simply waiting for the customers to validate the products and incorporate them into their systems. Based on these parameters, we can estimate our market share and identify potential design wins for the coming year.

Currently, Semiconductor International holds a market share of 25-30% in the smartphone market and 25% in the consumer goods market. By analyzing the market share in the industrial and automotive sectors, we can make predictions about the demand for the next year. For example, the demand for smartphones is expected to remain stable or increase by 1-2% compared to this year. We have accurate forecasts for the overall demand for smartphones, which are released in April and October. The same applies to accurate predictions for the automotive and PC industries. This depends on the overall industry conditions.

We can also accurately determine whether Semiconductor International's market share is increasing or decreasing within the industry. Therefore, as I mentioned earlier, the outlook for the entire year is relatively smooth and follows a W-shaped pattern.

The construction of our factories is not significantly affected by the economic conditions of the year. Even if the market conditions are favorable when we start building the factory, they may change in the following year. However, the factory will operate for 20 years. Therefore, the impact of a single year on the overall operation of the factory over 20 years is not significant.

My main focus is on the international market in the future, as it is crucial for Semiconductor International. Many international customers want a competitive manufacturer like Semiconductor International or a manufacturer with a significant market share in China. They prefer local production and require the production capacity that Semiconductor International can provide. We have set a target for the percentage of market share we want to achieve and the percentage of domestic demand we want to capture. Based on this, we have determined the amount of production capacity that Semiconductor International should build in the next few years. We have already announced the construction plans, including the number of wafers to be produced per month in each location. In the past few quarters and year and a half, we have announced these projects, and so far we have not changed our minds or felt that they are no longer needed.

The current downturn is just a preparation for the future. We can adjust the timing more cautiously, but the overall plan has not changed. Therefore, SMIC will still proceed with the construction plans that have been announced.

Competition among peers will always exist. As I mentioned earlier, SMIC has strategic partnerships with customers, and our platform needs to have high performance and full coverage. We also strive to make our costs highly competitive, so that we can provide customers with competitive prices in a timely manner. If we can achieve these three goals, SMIC can still maintain its advantages in the competition.

Q6: Regarding ASP prices, if the recovery is delayed for another year, the current wafer prices are relatively high compared to historical levels. I don't know if we should remain cautious about wafer prices in the coming quarters.

A6: As we have seen in the reports for the third quarter, many medium and small design companies have achieved design wins in smartphones and automobiles, and have achieved their best performance in terms of revenue and margin. As a supplier to these companies, SMIC's prices will remain stable and will not only increase.

However, for bulk products, in order to maintain our utilization rate with such large capacity, we need to follow the market. We are aware of the market price for bulk products such as CIS. If the current market price is, for example, 50 cents per chip, we cannot expect customers to place orders with SMIC at 55 cents. Therefore, bulk products will definitely follow the market, and we expect further declines in both volume and ASP for bulk products next year.

Q7: Regarding the capacity demand for our 40nm and 55nm nodes, is there still a tight situation? If so, what are the bottleneck machines related to this?

A: The 40nm node is still tight at the moment. There are two major requirements. One requirement is that for the current CIS (CMOS Image Sensor), everyone needs to do backside thinning and backside process. This involves more than ten to twenty steps of masks. Then, the logic circuit is bonded with copper, or even more advanced, two CIS chips are thinned, and then bonded with ISP (Image Signal Processor), making it a total of three chips. This process has a high demand for capacity, and the current capacity is not enough. This is probably the tightest situation in the world right now, because HBN (Hybrid Bonding Network) or Chiplets, or CIS Bonding, all require backside process, and it's not just one or two steps. It involves multiple layers of masks, about 15 to 20 layers. The capacity for backside process is very tight, and the suppliers have slow delivery, which is one of the bottlenecks. In the second aspect, there is currently a high demand for applications in the 40-nanometer process. For example, for MCU and HV DDIC, a strong and specialized ion implantation equipment is required. CIS is sensitive to contamination, so only certain types of ion implantation can be used while others cannot. Additionally, stress can be applied to certain layers of MCU, but not to others.

Therefore, currently, there are mainly specialized machines for producing DDIC and MCU. There are no issues with producing standard logic and IT components. Therefore, in the 40-nanometer process, attention should be paid to the backside of CIS, and the production capacity for MCU and DDIC is very tight.

The 55-nanometer process was originally used for the CIS sensor part, and another part was used for DDIC, which is relatively competitive and used in low-priced smartphones and displays. If the 55-nanometer process is further shrunk, it becomes very competitive at around 50 nanometers. Currently, the production capacity in this area is relatively insufficient.

In addition, the 55-nanometer process is also used for NOR flash. As I mentioned earlier, in this quarter, SMIC's NOR flash production increased by 27% because the industry reached a low point and prices were low. Companies started to stock up as the previous inventory was almost sold out. In addition to HBM and new PCs, memory products will see significant growth next year. When the price is lower than the production cost, many users, such as smartphone, automotive, and PC users, start to stock up and buy in large quantities. It is not possible for a product to always be produced below the production cost. If the manufacturer keeps losing money to produce it, they will gradually reduce production. Therefore, there is a high demand for the 55-nanometer process in these two areas.

Another application related to the 55-nanometer process is related to BCD and power supply. Power supplies require higher voltages, such as 14V and 5V, and have larger die sizes. With the increasing area requirements for GPUs, it becomes difficult to fit them in. Therefore, it is necessary to produce BCD analog power at the 55-nanometer and 65-nanometer nodes, with a die size of 1V or 1.4V and 1x1 size. This is also a significant demand. This area overlaps with the production of data center GPUs and requires not only GPUs and HBM, but also power supply chips with small die sizes. This is why there is currently a supply shortage in the 55/65-nanometer process. Q8: In this quarter, the proportion of our 8-inch wafer revenue has actually increased slightly. I would like to ask if this means that the situation for 8-inch analog and power management chips has improved? How do we judge the future demand for the 8-inch segment?

A: As mentioned earlier, there is currently a small boom in the Chinese smartphone market, and it is similar worldwide, but more pronounced in China. People are upgrading their phones, and new models are being launched. This has led to an increase in demand for smartphones, and as a result, other companies related to smartphones have started stocking up. As you know, these companies, such as those involved in fast charging, power management, Bluetooth, and Wi-Fi, all require 8-inch wafers, so the demand for them is relatively high. It has basically returned to the level of the peak period last year.

In the third quarter, more than 10 of these companies announced that they had reached a new historical high. These are the small and medium-sized design companies in China, many of which have already been listed on the A-share Science and Technology Innovation Board. This is the demand for 8-inch wafers from these companies. However, in terms of the overall estimate, as I mentioned earlier, this is just a small boom. If the overall average for next year is similar to the average for this year, then I am not particularly optimistic about the 8-inch segment. I think the loading ratio may remain at a relatively high level for quite some time, because the loading ratio for 8-inch wafers worldwide is currently around 50% to 70%. The better ones may be over 70%, while the poorer ones may only be around 50%. Even if there is a slight increase in demand, it may be taken away by low prices, so there won't be a significant improvement overall. This is different from what I just mentioned, such as backside processing of wafers or copper bonding, hybrid bonding, etc. Those are different because sudden large demand has emerged in niche markets, and capacity must be increased.

Q9: The additional capacity we released in the third quarter this year is relatively large. Can you break down the structure of the additional capacity, such as its size and applications?

A: At the beginning of the year, in February, we specifically discussed this year's plan, how much increase in 8-inch and 12-inch, and where the increase will be. The additional 8-inch capacity released in the third quarter is mainly in our Tianjin fab, and it is mainly for the production of 0.15 to 0.18 BCD Analog Power chips that I mentioned earlier. As for the additional 12-inch capacity, it is allocated to the 40nm and 55nm processes and comes from our Beijing and Shenzhen fabs. The 8-inch capacity will increase by 40,000 wafers for the whole year in Tianjin, including the fourth quarter, and the 12-inch capacity will increase by over 20,000 wafers for the whole year.

Dolphin Research's historical articles on SMIC:

Earnings Season

August 11, 2023 conference call: "The Surprising Increment of Smartphones: Is it from 'Trade-in' Programs? (SMIC 2Q23 Conference Call)" August 11, 2023 Earnings Report Review: "How long do we have to wait for the recovery of Semiconductor Manufacturing International Corporation (SMIC)?"

May 12, 2023 Conference Call: "Structural Recovery of Semiconductor Industry with 12-inch Rush Orders (SMIC 1Q23 Conference Call)"

May 11, 2023 Earnings Report Review: "SMIC: The Alpha Light Shining Through the Chip Cycle"

February 10, 2023 Conference Call: "High Depreciation Pressuring Gross Margin, Improvement Depends on the Second Half of the Year (SMIC 4Q22 Conference Call Summary)"

February 10, 2023 Earnings Report Review: "SMIC: Visible Decline, but Is Mediocrity Good Enough?"

November 11, 2022 Conference Call: "Capital Expenditure Unaffected Despite Semiconductor Downturn (SMIC 3Q22 Conference Call)"

November 11, 2022 Earnings Report Review: "SMIC: Long-Term Faith Cannot Escape the 'Cycle Curse'"

August 12, 2022: "How Will SMIC Respond to the Semiconductor Downturn? (2Q22 Conference Call Summary)"

August 11, 2022: "Unable to Raise Prices, SMIC Resists the 'Cycle Plunder'"

May 13, 2022 Conference Call: "Limited Impact of the Pandemic, Structural Shortage in the Semiconductor Industry (SMIC Conference Call Summary)"

May 12, 2022 Earnings Report Review: "Pandemic and Market Challenges? SMIC's Performance Remains Strong"

February 11, 2022 Conference Call: "SMIC Expands Production Beyond Industry Price Increases" Earnings Report Analysis on February 10, 2022: "Continuous Growth and Strong Performance of SMIC"

Telephone Conference on November 12, 2021: "Despite Exceeding Expectations, SMIC Experienced a Sharp Decline - What Did the Management Discuss?"

Earnings Report Analysis on November 11, 2021: "Stop Questioning Whether the Cycle Has Peaked - SMIC Still Going Strong!"

Telephone Conference on August 6, 2021: "How Does the Management View SMIC's 2021 Q2 Earnings Report?"

Earnings Report Analysis on August 5, 2021: "The Rising Force of China's Semiconductor Industry - SMIC"

In-Depth Analysis on December 29, 2022: "Semiconductor Avalanche? True Resilience Only After the Most Brutal Decline"

Industry In-Depth Analysis on June 24, 2022: "Order Cancellations - Is the Semiconductor Industry Really Going to Change?"

Company In-Depth Analysis on July 16, 2021: "SMIC (Part 2): The Underestimated Chinese Semiconductor Giant"

Company In-Depth Analysis on July 9, 2021: "SMIC (Part 1): The Strategy of the Leading 'Chip' Dragon"

Live Broadcast on May 13, 2022: "SMIC (00981.HK) Q1 2022 Earnings Conference Call"

Live Broadcast on February 11, 2022: "SMIC (00981.HK) Q4 2021 Earnings Conference Call" November 12, 2021, "Third Quarter 2021 Earnings Conference Call of Semiconductor Manufacturing International Corporation (00981.HK)".

August 6, 2021, "Second Quarter 2021 Earnings Conference Call of Semiconductor Manufacturing International Corporation (00981.HK)".

May 14, 2021, "First Quarter 2021 Earnings Conference Call of Semiconductor Manufacturing International Corporation (00981.HK)".

Risk disclosure and statement of this article: Dolphin Research Disclaimer and General Disclosure.