As of the end of June, the Shanghai Composite Index fell 5.37% from the beginning of 2022, the Shanghai and Shenzhen 300 fell 11.04%, the computer industry fell 23.69%, and the ChiNext Index fell 14.99%.

(Producer/Author: China Galaxy Securities, Wu Yanjing, Zou Wenqian, Li Luxin)

. Review of the market trends of the computer industry in the first half of the year

(I) The overall market performance of the computer industry in the first half of the year was poor, with a relatively clear bottom

due to multiple factors. In the whole of the first half of the year, the computer sector ranked first among all industries (wind Shenwan first-level industry range rise and fall, arithmetic average). As of the end of June, the Shanghai Composite Index fell 5.37% from 2022, the Shanghai and Shenzhen 300 fell 11.04%, the computer industry fell 23.69%, and the ChiNext Index fell 14.99%. At present, the overall valuation level of the computer industry (PE, TTM, excluding negative values) is 35.21 times, which is lower than the historical average of the industry in the past decade, and the industry already has allocation value.

The computer sub-sector has performed relatively well under the overall weakness of the overall weak sub-sectors include smart cars, smart transportation, smart cities, and cloud computing . At present, the overall valuation levels of the Shanghai and Shenzhen 300 and the computer industry (PE, TTM, excluding negative values) are 12.78 times and 35.21 times respectively. The computer industry is lower than the historical average of the industry in the past decade, and the industry already has certain allocation value.

2. In the first year of intelligent driving, the value of software has gradually increased

With the recent sales of many new energy vehicles, the cost of lidar has gradually approached a relatively reasonable price range. The mass loading of lidar required for autonomous driving at L3 level and above is the implementation of high computing power AI chips, and the first year of intelligent driving is truly here. Most of the new models represented by Huawei and new car-making forces have been equipped with L3-level autonomous driving systems. Huawei AITO's first new energy vehicle was officially delivered in March this year, and its sales volume exceeded 7,000 by June. It is expected that this year's sales will be 15,000 vehicles, catching up with the first echelon of new car-making forces. The value of automobiles' "intelligent empowerment" is gradually highlighting, and the value of software will gradually increase in the future.

(I) Recently, policies related to intelligent driving have been intensively issued

In the past year, policies related to intelligent driving have been intensively issued, and L3-L4-level models are expected to usher in large-scale catalysis. Recently, the legislative plan for 2022 issued by the Standing Committee of the Shenzhen People's Congress, the "Regulations on the Management of Intelligent Connected Vehicles in Shenzhen Special Economic Zone" (hereinafter referred to as the "Regulations on Administration") ranks first, and the relevant management regulations have passed the third review stage before the publication. This "Management Regulations" is the first official management document in the country to detail the rights, responsibilities, definitions, etc. of the L3 and above autonomous driving. It is expected to clear policy obstacles for vehicles that have achieved the legal status of autonomous driving on the road.

(II) L3-L4-level models are expected to be mass-produced in intensively this year and next year

From the mass production plan for major high-grade autonomous driving cars from 2021 to 2023, it can be seen that the models mass-produced in 2021 are mainly L2+, and a large number of L3+ models will be intensively produced in 2022, while the L4-level models are mainly planned to be mass-produced in 2023. Judging from the sales volume of models that have been mass-produced, Tesla and Xiaopeng rank at the forefront, with obvious advantages.

(III) Looking at investment opportunities from the intelligent driving industry chain

In this chapter, we aim to explore certain investment opportunities by discussing the structure and pattern of the industrial chain. Our core point of view: 1. Chips are the core of the industrial chain, and software manufacturers deeply bound to core chips benefit the most. 2. The operating system and application algorithm pattern is relatively complex. Tier1 is also actively competing with third-party suppliers at this level. The car manufacturer has a strong voice. Which cooperation model to choose is mainly determined by the car manufacturer. 3. In domain controllers, the value of autonomous driving domain controllers is very high, and there is huge potential development space. The main reason is that the demand for domain controllers for L4-L5-level autonomous driving will explode. However, due to the high cost of lidar, the large-scale mass production of L4-L5-level autonomous driving will still take some time.

1. The proportion of software value has greatly increased

In the process of the evolution of automotive E/E architecture from distributed to centralized, software is in an increasingly important position.First of all, domain controllers, software operating systems, middleware and algorithms are new links in the industrial chain. The cost of automotive software research and development will gradually increase and become the core of the value of the vehicle. It is expected that by 2030, the proportion of software costs in the vehicle's value will reach 50%; with the improvement of intelligent driving levels, the intelligent and diversified functions of automobiles in the future need to be equipped with corresponding software support, and the differences in the performance and function of the software will determine the differences in the automobile; enterprises in the automobile production industry chain will pay attention to and strengthen the construction of software capabilities to achieve internal changes from development model, organizational structure, personnel composition to operation system, etc. According to McKinsey's report, the global automotive software (algorithm, middleware and OS) market size will be approximately US$34 billion in 2020, and is expected to increase to US$63 billion in 2025 and US$84 billion in 2030, with a compound growth rate of 9% from 2020 to 2030. The annual compound growth rates of software market size in each domain are: OS and middleware +11%, body domain software +10%, power domain and chassis domain software +1%, smart cockpit domain software +9%, and autonomous driving domain software +11%.

2. The main links and patterns of the intelligent driving industry chain

domain controller, as the functional center of each domain, can realize the separation of software and hardware in each domain, unified standards, and greatly enhance the scalability of functions. The evolution of automotive E/E architecture from distributed to centralized is a major trend in industry certainty. According to the development stages of the six E/E architectures proposed by Bosch , most of them have developed to the stage of domain controller E/E architecture, which is to divide the automotive architecture into several regions, integrate and integrate the ECUs in each region, and control the functionally by the domain controllers in each region. In this way, the software and hardware in each domain can be separated, unified in standards, and greatly enhanced in function scalability. The smart cockpit domain and the autonomous driving domain are the focus of industry attention and competition. Currently, most Tier1s are divided into five major areas according to their functions, including smart cockpit domain, autonomous driving domain, power domain, chassis domain and body domain. The smart cockpit domain and the autonomous driving domain are the focus of industry attention and competition. The power domain and the chassis domain need to be combined with the functions of the autonomous driving domain and have a high technical threshold and are still in the early stages of development. The smart cockpit domain has developed from the central control system, and the autonomous driving domain is a new part. The two have added a large amount of software content in the integrated development, which tests the software capabilities of Tier1 suppliers.

smart cockpit chip and autonomous driving chip have a tendency to integrate, and the bottom layer can be universal. Nvidia Orin X's rich ecosystem and ADAS function implementation have high barriers and higher prices. Judging from the mass production plan next year, Nvidia Orin X has a market advantage. Car manufacturers are becoming more and more curious about the openness of the chip ecosystem, and the market acceptance of the black box solution of Mobileye has declined significantly. Among domain controllers, there is huge potential development space for autonomous driving domain controllers, mainly because the demand for domain controllers at L4-L5 levels will explode. According to McKinsey's report, the market size of ECU/domain controllers in different domains has an annual compound growth rate of: body domain +3%, power domain +5%, chassis domain +1%, smart cockpit domain -2%, autonomous driving domain (L1-L3) +3%, and autonomous driving domain (L4-L5) jumped from 0 in 2020 to US$34 billion in 2030; this is mainly based on the assumption that L4-L5-level cars are expected to produce 6 million vehicles per year by 2030, and the estimated production of L3-equipped cars will grow from the order of 100,000 in 2020 to more than 10 million vehicles in 2030. In the smart cockpit domain, the ECU/domain controller is mainly used to control the main dashboard, navigation system or vehicle audio system. The market size of smart cockpit domain ECU/domain controllers is expected to decline at a rate of about 2% per year, mainly because hardware costs are lower than those in distributed architectures, and although software costs have increased, the price of smart cockpit domain controllers is expected to drop further as the scale effect brought about by increased sales is gradually amortized. The pattern of operating system and application algorithms is relatively complex, and the industrial division of labor and cooperation model is still in the stage of continuous exploration.There are currently three industrial models: 1. The car manufacturer makes its own application algorithms, and Tier1 takes into account hardware production and operating system customization development; 2. The car manufacturer's application algorithms, operating system customization development and domain controllers are all made by third-party suppliers; 3. The car manufacturer completes everything from application algorithms to domain controllers by itself, similar to Tesla. In general, Tier1 is also actively competing with third-party suppliers at the software algorithm level. Car manufacturers have a strong voice. Which model to choose is mainly determined by car manufacturers. To a certain extent, it has also led the direction of combining software and hardware in the industry. We have sorted out the bicycle value and main participants in each link of the autonomous driving domain and the smart cockpit domain to see the development space and evolution of each link. The autonomous driving domain can be divided into perception layer, decision-making layer and execution layer according to the technical architecture. Among them, the main related to software are the perception layer and decision-making layer, and the execution layer is mostly related to automotive parts. The perception layer mainly includes a variety of on-board sensors, and the decision-making layer uses the domain controller as the functional center, integrating chips, software operating systems, middleware, and application algorithms.

As the intelligent driving level increases, the unit price of sensors of the perception layer tends to decline, but the number of use will increase. The L3 level requires 2 long-range lidars, and the L4-L5 level requires 4 long-range and 4 short-range lidars. The autonomous driving domain controller is a very high value link for bicycles, but we still have to wait for the L4-L5 level to increase the volume. The combination of

vision + lidar is the future development trend. There are two main routes for autonomous driving. One is the "visual school" represented by Tesla, that is, the vision-oriented solution dominated by cameras and is coupled with low-cost sensors such as mmWave radar. The other is the "lidar school" chosen by most autonomous driving manufacturers except Tesla, that is, the "lidar school" dominated by lidar , and is coupled with sensor components such as cameras and millimeter wave radars. The "visual school" is cheaper and has higher commercialization feasibility; the "lidar school" is currently costing higher, but it is more accurate in obtaining information. We believe that the combination of vision + lidar is the future development trend. The main logic is as follows: First, the pure vision solution has a flaw in recognition accuracy, and the real L4 cannot be achieved; second, the reason why lidar has not been widely used on a large scale is mainly because of its high price, but the price is expected to drop significantly in a few years; third, lidar cannot completely replace the camera function, and the L4 solution is likely to be a combination of vision + lidar. Overall, monocular camera solutions will still be the mainstream at this stage, and lidar will gradually increase with the cost of falling lidar. In the next few years, lidar will become an indispensable part of the L3-L5 autonomous driving system, and the combination of vision + lidar is the future development trend. The huge application potential of lidar will make it the fastest growing segment in sensors in the future. According to McCann Tsing's forecast, the market size of automotive sensor segments will have an annual compound growth rate of: lidar +80%, camera +7%, radar +13%, and others +6%. The market size of lidar will reach US$12 billion in 2030.

(IV) Huawei's car manufacturing ecosystem continues to grow, pay attention to Huawei's automobile industry chain

Huawei cooperative car industry M5 has delivered over 10,000 in 87 days, and the fastest record of single model payment of innovative brands exceeding 10,000. The M5 is the first model of AITO, a high-end brand created by Xiaokang Syris and Huawei. Recently, the May production and sales report released by Xiaokang Co., Ltd. showed that the sales of new energy vehicles reached 10,452 units in that month, and increased by 244.04% year-on-year. Among them, Cialis sold 5,440 vehicles, a year-on-year increase of 1,616.09%. In May, the production of new energy vehicles reached 11,077 units, a year-on-year increase of 273.97%, and the production of Selis was 5,650 units, a year-on-year increase of 1,082.01%. Among them, the M5 in Wenjie has delivered more than 10,000 units in 87 days, reaching 11,296 units, and the monthly delivery volume in May exceeded 5,000 units, and the fastest record of single model delivery of innovative brands exceeded 10,000. Huawei, which repeatedly stated that "don't build cars, it just helps car companies build good cars", has created a sales record of the fastest single model of new brand cars that exceed 10,000 yuan.Under the trend of decoupling of smart car hardware and hardware , the future computing power and software value are expected to be concentrated towards strong Tier1 such as Huawei, and the incremental hardware value will be left to the supply chain. At present, AITO Ask M5 is the first model equipped with HarmonyOS smart cockpit. It has now pushed the first OTA upgrade service, involving the optimization of many functions such as welcome seats, lighting prompts, smart voice, large-screen UI interface, turn signal prompt sound, etc. The HarmonyOS smart cockpit has built a rich window display form such as full screen, split screen, and floating windows to meet the interactive needs in different scenarios. (Report source: Future Think Tank)

(V) L4-L5 is expected to be the first to apply

We expect that L4-L5 will take several years to large-scale commercialization in complex scenarios, mainly because the cost of lidar is still not low. However, in relatively simple commercial vehicle scenarios, it can be achieved without lidar, and it will help achieve economic benefits, and it is expected to be the first to achieve large-scale applications.

Currently, there are seven major mainstream application scenarios for autonomous driving in the field of cargo commercial vehicles. According to the operating speed, it can be divided into high-speed scenarios and low-speed scenarios. Among them, high-speed operation scenarios include travel services and trunk logistics, and low-speed scenarios include mining site scenes, port scenes, airport scenes, logistics parks, and terminal logistics. The complexity of the scenario is one of the key factors in implementing autonomous driving technology in specific scenarios. Judging from the characteristics and indicators of scenarios such as traffic signs and pedestrian interference, it is less difficult to apply high-level autonomous driving technology in mining areas, ports and logistics parks, and is easier to achieve commercial application. In addition to being related to the complexity of the scenario, the large-scale implementation of applications also needs to test the business model, that is, the cost and profit of vehicle technology breakven point is very important. Labor costs continue to rise, while vehicle costs continue to fall. When the two reach the intersection, economic benefits can be generated for the company, and enterprises are willing to copy in large quantities. Currently, the cost of unmanned delivery vehicles ranges from 200,000 to 500,000 yuan depending on the sensor solution, and is expected to drop to less than 100,000 yuan in the next three years. When the cost of a whole vehicle is only 100,000, scenarios that are prone to revenue generation, such as fresh food, retail and express delivery, can be applied on a large scale. Therefore, it is expected that the closed low-speed loading scenario will take 2-3 years to land, the high-speed loading scenario will take 5-8 years, and the high-speed manned scenario will take more than 10 years to open. According to Marklines data, the global commercial vehicle sales reached 22.54 million in 2020. According to the 5-year replacement cycle, the global commercial vehicle ownership is about 100 million. We expect that after the high-level autonomous driving technology is more mature, based on economic benefits and safety considerations, the commercialization of commercial vehicle autonomous driving applications will be faster, and the penetration rate will gradually increase, and the commercial vehicle autonomous driving market space is relatively large.

3. Cloud computing is in full swing, and digital economy dock server is improving for a long time

(I) The growth rate of public cloud is high, but the margin is slowing down, and the growth rate of non-public cloud business has increased slightly

According to the latest report of iResearch Consulting , in 2021, China's overall IaaS market size reached 220.14 billion yuan, with a growth rate of 34.3%, and the PaaS market size reached 42.07 billion yuan, with a growth rate of about 55%. Overall, the slowdown in the growth rate of cloud computing IaaS market is due to the adjustment of the IaaS product structure, which has been tilted from the early stages of "face products" that can provide revenue growth such as CDN to computing power to match the larger basic demands of future digital economy construction. The PaaS market took over IaaS and gradually became the growth driver of the basic cloud market. From the perspective of product structure, databases and big data will still be the leader in the PaaS market growth in the short term, and artificial intelligence contributes weakly, but in the long run, AI is expected to become an incremental factor supporting the further development of the PaaS market.

In the non-public cloud field, the demand for government cloud and financial cloud is relatively prominent. Due to the certain adjustments in the government industry's cloud usage standards and cloud usage models in 2021, government cloud manufacturers have made necessary adjustments according to the new rules when providing relevant cloud services. Some projects have been delayed in construction, while financial cloud continues to grow, further increasing the market share. In addition, the expenditure on industrial cloud and medical cloud is increasing.

(II) The market structure of IaaS+PaaS tends to differentiate, and the growth rate of operators has risen

According to IDCh, the overall market size of China's public cloud services (IaaS/PaaS/SaaS) in 2021H2 reached US$15.13 billion (about RMB 102.1 billion), of which the IaaS+PaaS market in 2021H2 increased by 43.0% year-on-year.

Alibaba Cloud continues to rank first in China's public cloud IaaS+PaaS market share. In 2021H2, its market share is 36.7%, a decrease of 1.2% from 37.9% of 2021H1. Among them, the IaaS market share was 37.8% in the second half of the year, a decrease of 0.8% from 2021H1 (38.6%). The second-ranked public cloud IaaS+PaaS Tencent Cloud market share was 11.1%, a decrease of 0.1% from 2021H1 (11.2%). Among them, Tencent IaaS market ranked third, with its market share of 10.9%, a decrease of 0.2% from the first half of the year (11.1%). Huawei Cloud, the third largest public cloud IaaS+PaaS market, has a market share of 10.8%, a decrease of 0.1% from 2021H1 (10.9%). In its public cloud IaaS market, Huawei cloud ranked second, with a market share of 11.4%, an increase of 0.2% compared with 2021H1 (11.2%). It is worth mentioning that the operator's public cloud business is showing an accelerated trend, among which China Telecom 's Tianyi Cloud maintains the fourth place in the 2021H2 China Public Cloud IaaS+PaaS market, and its market share has increased from 8.3% in 2021H1 to 8.6%, which is the only one among the top five cloud manufacturers to expand its market share. Its public cloud IaaS market share has increased from 9.6% in 2021H1 to 10.3% in the second half of the year. The emergence of China Mobile is also the mobile cloud , which ranks seventh in the public cloud market share. Its IaaS+PaaS market share increased by 137.8% year-on-year, and increased by 24.9% month-on-month. The two growth data rank first among the top ten cloud manufacturers; , China Unicom , the revenue growth rate of public cloud IaaS, PaaS, and IaaS+PaaS in the fourth quarter of 2021 ranked first among all cloud manufacturers. The proportion of government affairs cloud and large enterprises to go to the cloud continues to increase, which is beneficial to cloud service providers of the state-owned assets system. According to the report of the Development Research Center of the State Council, the proportion of government agencies and large enterprises going to the cloud will reach 61% in 2023.

The country regards cloud computing as the underlying basic technology, proposes a number of support policies to promote industrial development. In April 2020, the National Development and Reform Commission and the Central Cyberspace Affairs Office encouraged the use of digital technologies such as big data, artificial intelligence, and cloud computing to play a role in emergency management, epidemic prevention and control, resource allocation, etc. With the strong promotion of policies, cloud computing technology and market are becoming increasingly mature, and enterprises' awareness and ability to go to the cloud are also increasing.

(III) The SaaS market is in full swing and will maintain a high prosperity in the long run.

The global SaaS industry development remains stable: According to Gartner data statistics, the global SaaS service market size reached US$102.1 billion in 2019. It is estimated that by 2022, the global SaaS service market size will reach US$138.3 billion, and the CAGR from 2019 to 2022 will be 11%. The development of the domestic SaaS industry is in an early stage of rapid growth, and the market size is expected to be close to US$10 billion in 2021: According to iResearch data, the scale of China's enterprise-level SaaS service market reached US$5.1 billion in 2019, a year-on-year increase of 40%. With the improvement of domestic companies' awareness of the SaaS industry and the urgent need for enterprises to transform into , it is estimated that in 2021, China's SaaS market will reach US$9.8 billion, accounting for 8.32% of the global market. There is still room for development in China's SaaS market and the potential market is huge.

Global SaaS investment and SaaS products are growing steadily. Global enterprises have been increasing in terms of SaaS usage and investment. In 2018, SaaS spending and application continued to grow rapidly among enterprises of all sizes, with the average global spending on SaaS was US$343,000, a year-on-year increase of 78%. The use and investment of SaaS products also have an economies of scale.According to Blissfully, in 2019, enterprises with more than 100 people used SaaS on average, and as the size of the enterprise grows, the number of SaaS products used also shows a linear growth characteristic. As the scale of the enterprise expands, companies prefer to use specialized applications rather than just adding complex functional modules.

SaaS segment segmentation is divided into general SaaS and vertical SaaS according to the scope of service objects. General SaaS is suitable for the entire industry and mainly uses general management tools and technical tools, including instant messaging, collaborative OA, financial management, human resources management and other services. Vertical SaaS serves specific types of industry customers and provides software services that are more targeted and closer to customers' business needs.

(IV) The growth rate of the digital economy has accelerated, and the digital economy base server market is improving for a long time

2022 The "14th Five-Year Plan for Digital Economy Development" was issued, and the development of the digital economy has risen to the national strategic level. The plan emphasizes that the digital economy is the main economic form after the agricultural economy and industrial economy, and takes data resources as the key factor. It also clearly states that by 2025, the added value of the core industry of the digital economy will account for 10% of GDP. We believe that digital infrastructure is the carrier of digital economy development, and with the support of national policies, related industries will usher in new development opportunities.

In the digital economy era, computing power has become the core engine that drives national economic growth. The IDC's "2021-2022 Global Computing Power Index Report" points out that computing power has a long-term and multiplier effect on the pull of economic growth: for every 1 point increase in the computing power index on average, the digital economy and GDP will grow by 3.5‰ and 1.8‰ respectively. Servers are the base of the digital economy and the overall IT system. As the underlying computing power support, they will usher in a burst of demand in the wave of digital and intelligent transformation in all industries.

East and West Calculation project is an important move in the digital economy strategy and will further strengthen the implementation of the digital economy related. East Data Western Calculation will guide the computing power demand in the eastern region to the western region in an orderly manner, promote the effective allocation of resources, and help improve the overall computing power resource level in the domestic region. In 2020, China's GDP growth rate was 3.0%, and the data economy growth rate reached 9.7%. The gap between the two shows a trend of widening. According to data from the National Development and Reform Commission, as of now, the scale of my country's data centers has reached 5 million standard racks, and the computing power has reached 130EFLOPS (130 billion floating-point operation per second). According to the Ministry of Industry and Information Technology, the average annual growth rate of national data center frame size will remain at around 20% by the end of 2023. We believe that during the 14th Five-Year Plan period, with the strong support of national policies, the digital economy will accelerate its development, and the construction of computing power infrastructure is expected to enter a new round of high prosperity cycle.

(V) The meta universe brings data traffic demand, edge computing and AI bring new data computing power growth points

Under the background of the meta universe, data traffic demand explodes, cloud computing, edge computing, artificial intelligence and other supporting technologies will usher in rapid development, and the computing power gap drives the growth of server demand. The metaverse is a new generation of Internet traffic generated by the transformation of network transmission (5G) + transformation of interaction methods (VR/AR) + transformation of computing power (AI chips such as GPUs) + content ecosystem (UGC). Historically, from 1G/2G/3G/4G to 5G/6G, from PC Internet, mobile Internet to metaverse, every generation of Internet changes can bring a new round of technological dividend release, giving birth to new computing power leaders (such as Intel , Qualcomm ), algorithm leaders (such as Microsoft , Google , byte interaction), terminal leaders (such as Nokia , Apple ) and corresponding business forms. 5G transmission technology realizes the key features of low latency in the metaverse and accelerates the implementation of the metaverse form. According to the latest data from the Ministry of Industry and Information Technology, my country has opened more than 1.425 million 5G base stations in 2021, and the number of 5G mobile terminal connections has reached 520 million. It is estimated that more than 600,000 new base stations will be built in 2022, and it is planned to reach 2 million by the end of the year. We believe that with the current large number of 5G base stations, the goal of building 600,000 new base stations in the next three years is relatively conservative and reasonable. It is estimated that by 2024, the construction of 5G base stations will exceed 3 million stations.

Edge computing and AI bring new growth points in data computing power. According to IDC data, the size of China's edge computing market is 17.26 billion yuan, a year-on-year increase of 16.3%; it is expected that the size of the edge computing market will reach 138.5 billion yuan in 2025, with a compound growth rate of 43.7% in the next five years. As a supplement to the cloud computing device side, edge computing products and data volume will show a huge growth trend in the future. We believe that massive computing and storage requests will promote the rapid popularization of edge computing and form an important branch of the cloud computing industry chain.

In terms of intelligent computing power, AI training and big data analysis have continuously improved the computing power requirements, promoting the rapid growth of AI servers. The development of artificial intelligence puts higher requirements on computing power. With the rapid development of artificial intelligence algorithms, more model training requires huge amounts of computing power support, and computing power is limited by the physical limitation of chip technology. Its technological improvement speed has lagged behind the algorithm model. Some models have approached the limit of AI's computing power, and the computing power gap will promote the server to increase the amount to make up for the demand. According to iResearch Consulting data, China's AI chip market size has grown rapidly, with the market size growth rate in 2020 being 57.0%, and the CAGR is expected to be 47% in the next five years. The market size of training chips and inference chips in 2020 is similar. In the next five years, the growth rate of the market size of the inference chips is expected to be greater than that of training chips. Judging from the technical architecture of AI chips, GPU is the most widely used accelerated computing chip thanks to its high density and high parallel computing advantages. It is expected that the mainstream architecture will still be dominated by GPUs in the next five years, and the FPGA and ASIC architectures will maintain rapid growth. (Report source: Future Think Tank)

(VI) The growth rate of the global server market is expected to slow down in the second half of the year, and China leads the global

Historically, Taiwan Xinhua Technology Aspeed's monthly revenue data can be used as a forward-looking indicator of the server industry's prosperity, and it has 2-3 months of forward-looking guidance for global server sales data. (1) BMC chips are also known as substrate management controllers. Each server needs to be equipped with at least one BMC control chip. Xinhua Technology is the leader in BMC chips, accounting for 70% of the market share; (2) Xinhua Technology's delivery cycle is 1-2 weeks, which is compared with the 2-3 delivery cycles of Intel chips, so it can reflect changes in the server market 2-3 months in advance. It can be seen that Xinhua Technology's monthly revenue growth rate has increased significantly since July 2021, and the prosperity in the first half of the year is still good, but the growth rate is expected to slow down in the second half of the year. After entering 2022, Xinhua Technology's revenue in January continued to maintain a year-on-year growth of 60%, and has rebounded by more than 100% from the trough in October 2020. In the short term, the prosperity is still better. However, according to TrendForce's research, the global server market's previous production plans in ODMs have begun to slowly cool down recently. With the significant improvement in the long and short material cycles, the material reserves of server motherboard suppliers have begun to slow down in the second quarter. At the same time, production in some ODM factories has been affected by the epidemic, and global server shipments in 2022 are expected to increase by 6.5% in Q3 and decline by 5.7% in Q4.

On the other hand, CPU is the most core component of the server, and currently 95% of the market share is occupied by Intel and AMD. Therefore, we can map the growth of server market demand through the chip business growth rate of Intel and AMD, the leading upstream CPU manufacturers. It can be found that although Intel and AMD's chip businesses fluctuated between quarters, they generally maintained steady growth. Starting from Q3 2021, Intel's chip business revenue growth rate began to recover, while AMD's server processor-related revenue has always maintained a high level, which is mutually verified with the growth rate recovery of BMC chips leading by quarter, reflecting the rising prosperity of the server market. The compound revenue growth rate of Intel's data center business from 2016 to 2021 was 8.4%, and the year-on-year growth rate of data center business in 2022Q1 was 8.45%. The decline in year-on-year growth rate was mainly due to intensifying competition. The CPU of ARM architecture is occupying the market share of traditional X86 architecture chips; AMD's compound revenue growth rate from 2016 to 2021 was 30.9%, and the year-on-year growth rate of revenue in 2022Q1 was 87.81%.

The launch of the new generation of server chips is coming, and the replacement cycle of superimposed servers is expected to stimulate the renewal demand of cloud manufacturers and government and enterprise markets this year. (1) Intel will launch the next generation of Xeon scalable processor Sapphire Rapids in the second half of 2022. It uses Intel 7 process technology (10nm Enhanced SuperFin) and supports the DDR5 memory standard for the first time. It is the most featured Xeon server processor to date. (2) NVIDIA's next-generation GPU Hopper is mainly aimed at high performance computing, artificial intelligence, etc. It will adopt a 5nm process, with up to 140 billion transistors and a core area of 900 square millimeters, making it the largest GPU ever. (3) NVIDIA plans to launch Ampere Next GPU and Bluefield-3 DPU in 2022. The Zen 4 Genoa, which AMD plans to launch, is scheduled to be mass-produced and launched in 2022. It supports DDR5 and PCIe 5.0. The 5nm Zen 4c is mainly aimed at cloud service load scenario optimization, slightly later than the Zen4 Genoa launch. The former has a maximum of 128 cores and the latter has a 96 cores.

(VII) Downstream demand Cloud manufacturers' capital expenditure growth rate remains high in the short term

From the perspective of downstream demand of servers, server demand is directly linked to capital expenditure of customers in cloud computing and Internet fields. Considering that the growth rate of server chips began to recover in the fourth quarter of 2021, we judge that global cloud manufacturers' capital expenditures are expected to further increase in 2022. In 2021, the capital expenditure growth rate of leading cloud manufacturers in North America remained high, an increase of 35.66% year-on-year. Among them, 1) Microsoft's capital expenditure is affected by seasonal fluctuations; 2) Google and Meta capital expenditure is mainly used to purchase equipment required for data centers such as servers, and the growth trend corresponds to the recovery of the server market.

In the domestic market, the total capital expenditure of Internet giants Baidu, Alibaba and Tencent showed a stabilization and recovery trend in the first half of the year, and it remains to be seen in the second half of the year. As of the fourth quarter of 2021, Baidu, Alibaba and Tencent's overall capital expenditure reached 25.151 billion yuan, a year-on-year increase of 50.76%. Among them, Baidu and Alibaba's growth rate was relatively fast, with capital expenditures of 4.237 billion yuan and 9.253 billion yuan, a year-on-year increase of 94.54% and 90.94% respectively; on the other hand, Tencent's capital expenditure returned to positive after negative growth for two consecutive quarters, with a year-on-year growth rate of 20.73%. We believe that the capital expenditure growth rate of domestic Internet giants has gradually stabilized, showing a rebound trend in the first half of the year, with great long-term growth potential, but whether there will be slowing down in the short term needs to be further observed in the second half of the year.

(8) With the promotion of industry information innovation, information innovation servers are expected to accelerate the deepening of

Party and government information innovation from electronic documents to e-government affairs, CPU replacement will gradually shift from PC to server, superimposed on the development of financial information innovation and accelerate, and the scale of information innovation will continue to expand. For the information and innovation industry, the country has proposed a "2+8+n" system, and gradually realizes independence and controllability in this order. Since 2013, the Party and government industries have taken the lead in carrying out replacement plans from the administrative office electronic document system, and are predicting that the replacement of electronic document system will be completed in 2023. As the Party and Government Information Innovation advances from electronic document to electronic government affairs, the demand for servers and related industrial chains will be further increased. Among the eight key industries, financial information innovation was the earliest and gradually accelerated, followed by the telecommunications industry; energy, transportation, aerospace, and medical care are also gradually advancing and piloting; the remaining n industries will be launched one after another in 2023. According to Yiou Think Tank, the scale of domestic information innovation industry has reached trillions of yuan, and will reach 1.33 trillion yuan in 2025, with an annual compound growth rate of 4.8%.

In recent years, national policies have continued to drive the development of information technology. "Domestic substitution" is the keyword of the policy, and breakthroughs in key technology fields are key tasks. According to the "Several Policies to Promote the High-Quality Development of the Integrated Circuit Industry and Software Industry in the New Era" issued by the State Council, China's chip self-sufficiency rate will reach 70% by 2025, while the current self-sufficiency rate of domestic chips is less than 30%. However, in recent years, domestic chips have caught up. The information innovation project was implemented from 2020 to 2021, and domestic servers have begun to be used in party and government agencies and some industries. The availability of domestic servers has been initially verified. And as domestic chip manufacturers and server manufacturers continue to change their products, product performance will continue to improve.At present, Haiguang and Huawei HiSilicon occupy the main share in the domestic server CPU market, and their products have been commercialized on a large scale, and there is huge room for market development in the future.

4. The market space for industry information innovation and innovation has been gradually implemented, and performance has accelerated its implementation

(I) The "bottleneck" technology continues to be supported by policies, and the development space for information innovation and innovation has broad

2020. Since 2020, my country's information innovation and innovation have entered the stage of large-scale implementation and promotion. According to the "2+8+N" system proposed by the state (i.e., the eight key industries of the Party, government + finance, telecommunications, electricity, oil, transportation, education, medical care, aerospace + comprehensive application of information and innovation products). According to the "2021 China Information Innovation and Innovation Ecological Report" of Haibi Research Institute, the CAGR is expected to reach 37.4% in five years. The development ideas of the information and innovation industry are gradually becoming clear. From 2020 to 2021, the information and innovation project was implemented, and domestic servers have begun to be applied in party and government organs and some industries. The availability has been initially verified. From the past, the state of scattered fighting alone has gradually become a new model of systematic, ecological and upper and lower industrial chain cooperation. At present, the Party and Government Information Creation has entered the stage of down-to-earth promotion of districts and counties. Industry information innovation has entered the first year of promotion since 21, and has begun to significantly promote and penetrate the financial and telecommunications industries in 22 years. Taking the financial industry as an example, starting from 2021, the replacement of information and innovation in the financial industry will gradually extend from the office OA system to the business system and core system. The promotion in 2022 will expand in breadth and depth, and it is expected that the pilot units will be expanded to more than 400. The information technology industry is based on the information technology industry. Through technological innovation, it builds an industrial ecosystem of domestic information technology from underlying infrastructure to upper-level application software: covering from underlying hardware (chips, computing, storage, etc.), basic software (operating system, middleware, database) to upper-level application software (industry application software and general application software), plus information security protection for the entire process.

. The Science and Technology Innovation Board is in the bottom range of valuation, and the market making system is implemented and rebounded

(I) In the past three years of the opening of the Science and Technology Innovation Board, relevant systems have been continuously improved

0. On May 13, 2022, the China Securities Regulatory Commission officially issued the "Pilot Regulations on the Stock Market Making Trading Business of Securities Companies" (hereinafter referred to as the "Mark Making Regulations"), which will introduce a stock market maker mechanism on the Science and Technology Innovation Board. On the same day, the Shanghai Stock Exchange issued the "Implementation Rules for the Market Making and Trading Business of the Science and Technology Innovation Board of Stocks (Draft for Comments)" and the "Shanghai Stock Exchange Securities Exchange Business Guide No. X - Market Making for the Science and Technology Innovation Board (Draft for Comments)", and publicly solicited opinions from the market from May 13 to May 28.

(II) The market maker system is conducive to improving market liquidity and enhancing market stability

The market maker system is conducive to improving market liquidity and trading activity, promoting price discovery, and enhancing market stability. Market merchandising and bidding trading are the two most basic trading models in the financial market. Among them, the market maker system is that one or more market makers are responsible for providing bilateral quotations for buying and selling, and the investor's trading instructions will be transmitted to the market maker and traded with it; the bidding trading system, that is, the transaction is completed under centralized matchmaking of the exchange, and the investor buys a certain stock from another investor at the lowest price reported in the market and sells the stock at the highest price reported in the market. The difference between the two is mainly reflected in: (1) Market liquidity: Under the bidding trading system, investors' entrustment orders are matched and matched according to the trading rules, thereby forming price and liquidity. When the demands of both buyers are unbalanced or bulk transactions occur, the trading needs of some investors will not be met. Market makers are responsible for providing continuous bilateral quotations and can be used as counterparty to the transaction at any time to provide liquidity to the market; (2) Market stability: Under the bidding system, transaction prices generally fluctuate continuously with the changes in the power of buyers and sellers. If the power difference between the two parties is too large, the price will fluctuate abnormally.When the buying and selling power is uneven, market makers can accept the buy or sell orders, alleviate the imbalance of buying and selling instructions, and then curb the corresponding price fluctuations, which will help enhance market stability; (3) Transaction costs: Under the bidding system, investors directly conduct pairing transactions; and under the market makers act as trading parties to provide bilateral quotations, and the bid and offer price difference they obtain corresponds to the increase in the trading costs of investors. (4) Information transparency: In the continuous trading bidding stage, domestic Shanghai and Shenzhen securities companies publish the trading prices and order volumes in real time, and the market transparency is relatively high; and under the market maker system, the prices of securities are reported by market makers, and investors cannot know the market supply and demand situation in real time, and the market transparency is relatively low.

(III) The market making system continues to explore, and the areas involved are constantly enriched

The market making system continues to explore, and the areas involved are constantly expanding. The introduction of the market making system injects liquidity into various asset fields. At present, my country's New Third Board, Science and Technology Innovation Board and other companies have gradually introduced the market-making system, but it is still in its early stages of development. In other fields, my country's ETF funds introduced market makers, and the liquidity of ETFs introduced by market makers is significantly higher; 2 to 8 securities companies in the first batch of public REITs are responsible for market makers, providing guarantees for their liquidity; the Shanghai and Shenzhen Stock Exchange announced in March 2022 that China's depositary receipts will introduce competitive market makers under the bidding trading system. The market-making system injects liquidity into the New Third Board market. Before August 2014, the stock transfer model of the New Third Board was an agreement transfer. At this stage, the transaction amount of the New Third Board stocks in my country was at a relatively low level, the turnover rate hovered around 5%, and the liquidity of the New Third Board was relatively low. In August 2014, the New Third Board implemented a market-making system, with the purpose of providing liquidity to the New Third Board stock trading market. After the market maker system was applied to the New Third Board, the transaction volume and transaction amount of the New Third Board have greatly increased. Among them, the transaction amount of market making in 2014 accounted for more than 95% of the transaction amount of the New Third Board for the whole year, and the transaction amount far exceeded the transaction amount transferred by agreement. The average turnover rate for the whole year exceeded 20%, which was significantly higher than in 2012 and 2013. It shows the significant role of the market maker system in increasing market transaction volume. The scale of

ETF products has expanded rapidly, and the ETF liquidity introduced to market makers has been significantly higher. The scale of ETF funds in my country has expanded rapidly, with net asset value rising from 431.305 billion yuan in April 2018 to 1.374.600 billion yuan in April 2021, with an annual compound growth rate of up to 47.16%. On the other hand, liquidity is one of the core considerations for investors to purchase ETF products. Inadequate liquidity will not only lead to high impact costs of ETF trading and affect investors' trading experience, but may also directly lead to the failure of ETF investment. In May 2012, the Shanghai Stock Exchange issued the "Shanghai Stock Exchange Securities Investment Fund Liquidity Service Business Guidelines", and the Shanghai Stock Exchange issued the "Shanghai Stock Exchange Trading Open-end Index Fund Liquidity Service Business Guidelines", officially introducing the "market maker system" to the domestic ETF market. Market makers are conducive to improving ETF liquidity, curbing large price fluctuations, and bringing investors a better trading experience. As of May 17, 2022, 620 of the 699 ETFs on the market had introduced market makers, accounting for nearly 90%. Among them, the average turnover rate of ETFs introduced into market makers was 7.64%, while the average turnover rate of ETFs not introduced into market makers was only 1.54%, which shows that market makers have improved liquidity significantly.

market maker system provides guarantees for the liquidity of public REITs. Public REITs refer to standardized financial products that hold infrastructure projects through special purpose carriers such as infrastructure asset-backed securities, and use the cash flow generated by infrastructure projects as the source of repayment, and allocate most of the profits to investors. Referring to overseas experience, the turnover rate of REITs is relatively low, the liquidity in the secondary market is weaker than that of stocks, and investors may not be able to trade in time when buying or selling. In order to ensure the smooth operation of public REITs listed, the first batch of public REITs in China need to select no less than one market maker to provide bilateral quotations and other services to fully ensure the liquidity of REITs after listing.

(IV) There are 26 securities companies in the first batch that meet the market making conditions. The market making business mainly includes leading securities companies

market making business requires higher comprehensive business capabilities, and leading securities companies are expected to become the first batch of participants. The qualification conditions for the pilot project of the Science and Technology Innovation Board include: 1) The net capital in the last 12 months shall not be less than 10 billion yuan; 2) Classification ratings of A (inclusive) or above in the last 3 years. 26 listed brokers have met the criteria for the classification ratings and the net capital (or net capital by the end of 2020) at the end of 2021 (or net capital by the end of 2020), most of which are the leading brokers in the industry. The policy clearly states that securities firms can use their own stocks, stocks borrowed from China Securities Finance Corporation or other stocks entitled to be disciplinary as sources of market making bonds, and leading securities firms with abundant securities pools will benefit more. (Report source: Future Think Tank)

(V) The valuation level of the Science and Technology Innovation Board continues to decline, with good overall liquidity and growth ranking first in all sectors

The Science and Technology Innovation Board will usher in a three-year lifting period for major non-issues this year. In addition to the Fed's expectation of interest rate hikes and the impact of external situations, the sector's valuation continues to decline, and the high valuation is gradually digesting. As of June 2022, the P/E ratio of the 50 Index of the Science and Technology Innovation Board was 38 times (excluding negative values), which is far lower than the valuation average since the opening of the Science and Technology Innovation Board, and is at the bottom of the historical lowest value and has certain allocation value.

html From the beginning of 0 years to the present, the turnover rate of the Science and Technology Innovation Board's daily circulation market value has been 2.45%, slightly higher than the 2.08% of the A-shares on the Shanghai Stock Exchange Main Board. The average daily turnover of individual stocks is about 95 million yuan, and the total turnover of the Science and Technology Innovation Board accounts for 4.14% of the total daily turnover of the total A-shares.

According to the 2021 annual report data and the 2022 Q1 data, the growth of the Science and Technology Innovation Board is far better than other sectors. Horizontal comparison, according to the 2021 annual report data, in terms of operating income growth rate, the Science and Technology Innovation Board (37%), the Beijing Stock Exchange (31.7%), the ChiNext (23.2%), the main board (17.5%); in terms of net profit attributable to shareholders, the Science and Technology Innovation Board (62.9%), the ChiNext (25.9%), the Beijing Stock Exchange (23.7%), the Main Board (22.5%). According to the data on Q1 2022, in terms of operating income growth rate, the Science and Technology Innovation Board (45.8%), the Beijing Stock Exchange (32.8%), the ChiNext (20.4%), the main board (17.5%), and the net profit attributable to shareholders, the Science and Technology Innovation Board (62.5%), the Beijing Stock Exchange (26.5%), the main board (5.2%), the ChiNext (-14.7%). In the future, under the background of normalization of registration system IPOs, the market capacity of the Science and Technology Innovation Board will continue to expand, and the liquidity tension of some individual stocks will gradually emerge. The market maker system will help increase the liquidity of individual stocks and curb large price fluctuations.

(VI) Computer industry company indicators in the Science and Technology Innovation Board

According to the first-level industry classification of Shenwan, there are a total of 44 computer industry companies on the Science and Technology Innovation Board. Among them, 12 companies had revenue scale of more than 1 billion yuan. According to the 2021 annual report data, 34 companies had revenue growth, 22 companies had revenue growth rate greater than 20%, and 15 companies had net profit growth rate greater than 20%. The average R&D expense ratio of computer companies in the sector is 20.6%, far higher than the overall level of the computer sector.

6. Analysis of key companies

(I) Guoneng Rising: a leading enterprise in the new energy power generation power prediction system, multi-point layout opens up growth space

is a leading enterprise in the new energy power generation power prediction market. The company mainly provides new energy information products and related services to new energy power market entities such as new energy power stations, power generation groups and power grid companies, with new energy power generation power forecast products as the core, new energy grid-connected intelligent control systems, new energy power station intelligent operation systems, and power grid new energy management systems as the expansion. In 2019, the company's market share in the photovoltaic power forecast market and the wind power forecast market was 22.10% and 18.80% respectively, ranking first. The company has high-precision and high-stability power prediction technology, and its prediction accuracy is at the forefront of the industry. The company attaches importance to services and has formed a business system with providing services as its core. A service system with wide coverage and timely response has become an important advantage for the company over its competitors. It is expected that the company will continue to maintain its dominant position in the industry in the future. Power prediction systems are necessary for new energy power plants, and their accuracy and service importance will continue to increase. With the continuous expansion of the grid-connected installed capacity of new energy power stations, grid balance puts forward high requirements for the accuracy of new energy power generation power prediction.The assessment of power generation forecasts by energy regulators in various regions on new energy power plants is becoming increasingly strict. The accuracy of power generation forecasting will directly affect the operation and profitability of power plants, and the high accuracy of power forecasting needs to be achieved through high-quality services. Therefore, when selecting suppliers, downstream power station customers will pay attention to the accuracy of power forecasts, rapid feedback during the service period, and timely response.

During the 14th Five-Year Plan period, my country's average annual installed capacity will double, ensuring the high growth rate of the company's main business. The change in the company's revenue growth rate is consistent with the change direction of the scale of new photovoltaic wind power installed capacity in my country, and is highly correlated. During the 14th Five-Year Plan period, it is expected that the annual installed capacity of photovoltaic and wind power in my country will be about 120-140GW, while the average annual capacity before 2020 is only 51GW. The industry's annual new installed capacity doubles, which will drive the company's revenue to continue to increase in its power forecasting equipment and services, grid-connected intelligent control systems and other products. Technology extension, layout new energy information scenarios in multiple points to open up space for growth. 1. Power transaction auxiliary decision support System: The proportion of electricity in my country's market-oriented transactions has increased significantly, and the increase in the complexity of electricity pricing has led to the demand for power transaction auxiliary quotation decision software. The company's power transaction auxiliary decision-making support platform aims to help new energy power generation companies improve their power sales revenue and market competitiveness, and assist in participating in market-oriented power transactions. With the core of output forecasting, market forecasting and quoting, and volume-reporting auxiliary decision-making, we provide the overall declaration suggestions for spot power trading, medium- and long-term trading, and auxiliary service trading for new energy power generation groups and new energy stations. 2. Virtual power plant smart operation and management system: In 2021, the National Energy Administration revised the "Regulations on the Grid-connected Operation Management of Electric Power" and the "Regulations on the Management of Electric Power Auxiliary Services", encouraging new energy storage, adjustable load and other grid-connected entities to participate in power auxiliary services, and the demand for virtual power plant management platforms is generated accordingly. The company's smart operation and management system of virtual power plants can aggregate loads and store resources, participate in the power market, provide a variety of demand-side response services to the market, assist in stabilizing power balance, and assist in the consumption of new energy.

(II) Jiahe Meikang: Leading the electronic medical record industry, data centers and specialized electronic disease history are expected to become the second growth pole

electronic medical record rapid popularization + stock upgrade, driving the information construction of the entire hospital. The policy continues to promote, and there is still a lot of room for upgrading electronic medical records: 1. The three-year DRG/DIP action plan is released, and "full coverage" is achieved by the end of 2025. The implementation of DRG/DIP requires that the application level of the electronic medical record system must reach the level 5 standard; 2. The list of "Thousand County Projects" is determined, and 1,233 county-level hospitals within the scope are required to meet the third-level hospital standards. Assuming that the average investment in medical IT transformation of a single county hospital is 20 million, the market space of the "Thousand County Projects" will reach 24.6 billion during the 14th Five-Year Plan. Further, electronic medical records above level 5 will drive the growth of demand for medical data centers. my country's medical data center market is still small at present and is expected to accelerate growth in the next few years. Electronic medical records and the information construction of the entire hospital have opened up a large amount of upgrades. We calculated that the tertiary hospitals and secondary hospitals will contribute 601 and 48.2 billion yuan of market growth in 2022-2025 respectively. Together with the 24.6 billion yuan of increments of Qianxian Project, the total market space growth will reach 132.9 billion yuan from 2022-2025.

is a pioneer in clinical informatization and a leader in the field of electronic medical records. The company's electronic medical record business has firmly ranked as the leader in the industry. The company has rich product layout, with comprehensive electronic medical records as the core, horizontally expanding the product lines of specialized electronic medical records such as mobile nursing, hand anesthesia, ICU, emergency, oral specialty, obstetrics and other specialty electronic medical records, and vertically building a hospital data center, realizing the transformation from a single product model dominated by electronic medical records systems to a comprehensive model of overall solutions for smart hospitals. Drawing on the development history of the American medical IT industry, the advantages of technology-leading companies are constantly expanding. In the future, electronic medical records will develop towards specialization and intelligence. The specialized electronic medical records market space is expected to open up, and its structural proportion in the overall electronic medical record market will continue to increase. The company is in the industry leading position in the hospital data center and specialty electronic medical record market, and data center and specialty electronic medical record are expected to become the second growth pole.The growth rate of multiple businesses is impressive, and the scale effect is gradually reflected. In recent years, the company's operating income has grown rapidly, with a high degree of productization, and its gross profit margin is higher than the average of comparable companies, showing a long-term upward trend. The company's R&D expense ratio has been significantly higher than the industry average for many years. The level of sales and management expense control is increasing year by year, the rapid expansion of revenue scale is expected to increase efficiency and reduce fees, and profitability is expected to continue to improve.

(This article is for reference only and does not represent any of our investment advice. If you need to use relevant information, please refer to the original text of the report.)

selected report source: [Future Think Tank]. Future Think Tank - Official Website