laitimes

Special Report on Artificial Intelligence Industry: The Apocalypse of Historical Experience on the β of AI Industry

author:Think Tank of the Future

(Report producer/author: Northeast Securities, Deng Lijun, Xie Lixin)

1. Sufficient conditions for industrial switching

1.1. Historical industry switching

Since the 90s, the mainland's economic development has experienced several rounds of industrial switching, mainly including labor-intensive light industry (1992-2003), capital-intensive heavy industry (1999-2013) and real estate (1999-2016), technology-intensive mobile Internet (2010-2016) and new energy (2018-2022).

Industrial switching has shifted from labor-intensive to capital-intensive to technology-intensive now. Specifically: (1) Labor-intensive light industry (1992-2003): Since the reform and opening up, China has attracted foreign investment with cheap labor, formed a labor-intensive light industry system based on textiles and garments, and driven rapid economic growth. (2) Capital-intensive heavy industry (1999-2013): with the increase in household consumption income bringing about the demand for durable goods, including large household appliances and automobiles, the proportion of heavy industry in industry continued to increase; After entering the WTO, it will further promote the import and export trade of the manufacturing industry; After the "four trillion yuan" stimulus, the introduction of "sending automobiles and home appliances to the countryside" to further promote the sales of automobiles and household appliances. (3) Capital-intensive real estate (2003-2016): In 2003, the State Council positioned real estate as a pillar industry, and it can be seen that every round of economic fluctuations in 2005-2016 is inseparable from the pull of real estate; After 2016, it was proposed that "housing is not speculation" to gradually reduce the economy's dependence on real estate. (4) Technology-intensive mobile Internet (2010-2016): the rapid development of smart phones in 2010, the 18th National Congress in 2012 listed "implementation of innovation-driven development strategy" as the main direction, and mobile Internet with the popularization of mobile phones and 3G has become one of the leading directions of emerging industries; With the slowdown of Moore's Law, the user scale is relatively saturated, and the development of the mobile Internet after 2016 is slightly weak. (5) Technology-intensive new energy (2018-2022): Since 2017, domestic new energy policies have been frequent, the development direction of pure electric passenger vehicles has been determined, the traditional energy structure dominated by coal has been adjusted, and the penetration rate of new energy vehicles has continued to increase under the promotion of policies, and with the further determination of the "dual carbon" goal in 2020, the rapid development of new energy-related industries.

Special Report on Artificial Intelligence Industry: The Apocalypse of Historical Experience on the β of AI Industry

1.2. Conditions for industry switching

Looking back at history, we believe that there are three sufficient conditions for industrial switching: (1) there is endogenous resistance to the development of the previous industry; (2) the necessity of economic development and the top-level design of policies; (3) Basic elements necessary for industrial development. Since the macro environment of the previous rounds of industrial switching is quite different from the current one, and the current priority is high-quality development, the new industry will still be mainly technology-intensive, so we will specifically discuss the last two rounds, namely mobile Internet and new energy.

1.2.1. Mobile Internet (2010-2016)

The growth rate of real estate + exports has declined, and total factor productivity has slowed down. (1) The real estate growth center has moved downward. Since 2003, the growth rate of real estate sales has been downward, due to the continuous high growth of commercial housing sales, and the base has continued to rise; Stimulated by the "four trillion yuan," the sales of commercial housing also showed a high growth rate, but then fell rapidly, and at this time, housing prices rose too quickly and the economic structure was unbalanced. (2) The global economy is slowing down, and the growth rate of exports is declining. From the perspective of the external environment, under the impact of the economic crisis and the European debt crisis, external demand-oriented exports have been significantly dragged down; At the same time, the downward trend of the global economic growth center brings the downward trend of the export side. (3) Total factor productivity has slowed down, and structural adjustment is on the string. In the context of weak global demand, there is a need to improve the export-led, investment-led economic model, as well as the problem of low-end overcapacity; After 2008, total factor productivity slowed down, and even fell slightly around 2011, so the three consecutive Central Economic Work Conferences from 2012 to 2014 clearly called for "comprehensively resolving overcapacity" and "stabilizing growth, adjusting structure, and promoting reform".

The focus of policy has gradually shifted to "adjusting the structure". Since 2012, the government's economic center of gravity has gradually shifted from total to structure, and the 18th National Congress has made it clear that innovation driving is the leading strategy to accelerate the transformation of economic development mode, and technological innovation is the main direction of attack for cultivating and developing strategic emerging industries. Subsequently, relevant documents on "scientific and technological innovation", "technological upgrading" and "emerging industries" were issued successively, and promoting scientific and technological innovation became the key work of major ministries and commissions in 2013.

3G networks and mobile phones have built the foundation for explosive growth, and the scale of users has grown rapidly to achieve monetization. (1) The construction of 3G mobile networks has broadened the bandwidth bottleneck. On January 7, 2009, the Ministry of Industry and Information Technology officially approved the third-generation mobile communication (3G) business license of China's three major telecom operators, and China Mobile's network has fully entered the 3G era, and the mobile network speed has been greatly improved. (2) The rapid spread of smartphones. In 2012, traditional mobile phone manufacturers represented by Samsung and HTC have followed the Apple model and launched touchscreen smartphones; The emergence of smart phones and the acceleration of the network have given birth to rich application software of mobile smart terminals, and mobile Internet applications have shown explosive growth. (3) The rapid growth of user scale has become the basis for monetization: by the end of December 2012, the size of mainland mobile Internet users was 420 million, accounting for 74.5% of the total Internet users; Among them, the scale of smartphone Internet users reached 330 million, accounting for 79.0% of mobile Internet users, becoming an important carrier for the development of mobile Internet in mainland China, and the huge user group contains huge business potential.

1.2.2. New Energy (2018-2022)

The development of mobile Internet is limited by technological limitations and saturated penetration. (1) Moore's Law slowdown is subject to technical limitations. The benefits of hardware improvements are offset by larger operating systems and applications, and the skyrocketing cost of chip processes after 28nm and the difficulty of upgrading new processes have slowed down the performance of smart devices. (2) The scale of users tends to be saturated, and the development of each track slows down. Since 2018, the overall user scale of the mobile Internet has slowed down, reaching 1.37 billion in September 2018, the growth rate has decreased by more than 50% year-on-year, and the mobile Internet dividend has basically become saturated and the development has slowed down.

Special Report on Artificial Intelligence Industry: The Apocalypse of Historical Experience on the β of AI Industry

Dual carbon + energy security accelerates the development of new energy. (1) Promote the development of renewable energy and accelerate the transformation of the energy structure with dual carbon. Since 2017, the government work reports of the two sessions have attached great importance to the development of renewable energy, repeatedly involving "energy saving and emission reduction", "optimizing the energy structure" and "developing renewable energy", raising the transformation of the energy structure to the height of the national strategy; In 2020, the "dual carbon" goal was proposed, in order to achieve carbon neutrality, clean electricity will become the allocation center of the energy system in the future, and the supply side will be mainly photovoltaic + wind power; The demand side is fully electrified, supplemented by hydrogen energy. (2) The pattern of tight energy supply and demand remains unchanged. After the supply-side reform, the backward production capacity was cleared, and the energy supply and demand pattern was tight; The Russia-Ukraine conflict has further increased its importance to energy security, and the 2022 government work report of the two sessions proposed to fully ensure energy security. (3) The implementation of the double credit policy and the long-term mechanism of new energy vehicles have been strengthened: On September 28, 2017, the Ministry of Industry and Information Technology issued the "Measures for the Parallel Management of Average Fuel Consumption of Passenger Car Enterprises and New Energy Vehicle Credits", which drives traditional car companies to transform the layout of electric vehicles and promotes the new energy vehicle industry to enter a long-term growth stage.

The outbreak of the new energy industry has benefited from technological upgrading and early demonstration and promotion. (1) Supply-side reform clears backward production capacity. The supply-side reform aims to clear backward production capacity and then make up for shortcomings, and by 2018, the coal industry had resolved 690 million tons of excess capacity, completing 87% of the target of 800 million tons. It further provides a development foundation for promoting the transformation of the energy structure. (2) The improvement of lithium battery technology promotes the development of the new energy vehicle industry. On the one hand, the energy density of power batteries continues to increase, which greatly solves the anxiety of mileage, and the energy density of BEV passenger car battery packs reached 145.26Wh/kg at the end of 2019, and the cruising range continued to increase; On the other hand, salt lake lithium extraction technology has made a breakthrough, and in the past 17 years, salt lakes have begun to produce 7,000 tons of industrial-grade lithium carbonate, and gradually developed to battery level, lithium mining costs have declined, and the economy of electric vehicles has gradually been reflected. (3) Demonstration promotion and explosive models in the early stage promote the penetration rate to increase. On the one hand, the new energy vehicle market started early, experienced a demonstration and promotion period, a promotion and application period and a policy-driven period, and vehicle sales increased from 82,000 units to 6.887 million units from 2011 to 2022, and the proportion of sales increased from 0.04% to 25.63%; On the other hand, Tesla's launch of the Model 3 in 2017 became a hit, driving domestic car companies and new forces in car manufacturing to invest in pure electric research, driving the rapid increase in the penetration rate of new energy vehicles.

1.3 Whether the digital economy of .AI+ is an industrial switch

The growth rate of new energy vehicle sales has declined, and the problem of overcapacity on the supply side has become prominent. (1) In terms of new energy vehicles, in 2022, stimulated by a series of stable growth and consumption promotion policies such as halving the purchase tax, sales of new energy vehicles reached 6.887 million units, a year-on-year increase of 93.4%, and the market share reached 25.6%, an increase of 12.1% compared with 2021, and the penetration rate of new energy passenger vehicles in 2022 reached 27.6%, to a certain extent, overdraft the automobile consumption demand in 2023, and it is expected that the sales of new energy vehicles in 2023 will be 9 million units, a year-on-year growth rate of 35%. (2) Upstream, lithium carbonate and polysilicon production capacity has been released, and the price has fallen by more than 60% from the high level in 2022, effectively alleviating the cost pressure in the middle and downstream while shifting the development direction of the entire new energy industry from quantitative development oriented by raw material shortage to high-quality development in pursuit of cost-effectiveness, and profit space has declined; (3) In the midstream, in terms of new energy vehicles, by 2025, the power battery production capacity required for new energy vehicles will reach 1200GWh, but the current production capacity planning of power batteries has exceeded 5000GWh; In terms of new energy power generation, the domestic wafer production capacity is expected to reach about 740GW in 2023Q1, which is obviously overcapacity for upstream polysilicon and downstream cells and modules, and the industry competition pattern has deteriorated.

National ministries + basic systems + industrial policies catalyze to promote the development of AI+ digital economy. (1) Establishment of relevant ministries at the national level. After the two sessions, the State Council adjusted its institutions and established a new State Data Bureau, indicating that digital China, as a new strategy for the development of national informatization in the new era, has entered the stage of implementation by specific departments from strategy implementation; At the same time, the Department of Science and Technology was reorganized, mainly for the Ministry of Science and Technology to delineate part of its functions to focus on scientific and technological development and strengthen scientific and technological innovation; Subsequently, the Artificial Intelligence Planning and Promotion Office was established, and a new generation of major scientific and technological projects of artificial intelligence was launched, and corresponding deployments were made in digital twins, digital manufacturing, and smart medical care. (2) On December 19, 2022, the State Council issued the "Opinions on Building a Data Infrastructure System to Better Play the Role of Data Elements", putting forward opinions on issues such as property rights, transactions, circulation, and income distribution involved in data transactions; Standardize the development of data-related industries at the institutional level. On April 11, 2023, the Cyberspace Administration of China issued the Measures for the Administration of Generative Artificial Intelligence Services (Draft for Comments), which aims to standardize the development direction of artificial intelligence and achieve the benign development of artificial intelligence from the institutional perspective. (3) The "Overall Layout Plan for Digital China Construction" released in February 2023 clearly indicates the importance of digital China construction, and the deep integration of digital economy and real economy will be an irreversible trend of the times. Up to now, the central and local governments of the mainland have successively issued a number of policies related to the digital economy, and keywords such as "digital government", "promoting the digital transformation of small and medium-sized enterprises", "digital village" and "data security" have become the focus of policies.

The development of this round of digital economy is based on high-quality elements, computing power infrastructure and large models. (1) Scarcity of high-quality data elements: With the spread of AI technology application scenarios, data elements are further highlighted as the basis for large model training. As far as the domestic data market is concerned, according to the official disclosure of the National Development and Reform Commission, mainland government data resources account for more than 3/4 of the country's data resources, but the scale of openness is less than 10% of the United States, and the scale that individuals and enterprises can use is less than 7% of the United States. (2) Computing power infrastructure needs to be improved: On the one hand, the development of large models requires extremely high training computing power, and ChatGPT needs 10,000 NVIDIA A100 chips as the basis of computing power considering the loss of interconnection. At present, the gap between domestic large models and foreign countries in terms of computing power is too large, and the AI industry leaders mainly use NVIDIA's GPU chips for development and training, and the limited research and development capabilities of domestic high-performance GPUs restrict the development of algorithms and the improvement of large model training accuracy; On the other hand, as the main bearer of computing power, the demand for supporting data centers has also increased. (3) The AI large model has made a breakthrough, but it is still in the early stages of landing. On June 11, 2018, OpenAI released the GPT-1 model, and after 5 years of iterative development, the large-scale multimodal model GPT-4 was officially released on March 15, 2023, which can not only read text, but also recognize images and generate text results. At present, Baidu and iFLYTEK have launched the Wenxin model and the iFLYTEK Spark model respectively, and Tencent, Huawei, Ali and other accelerated layouts.

Compared with the development of smartphones, the landing of large models is expected to land within a 1-year cycle. In general, the current development of AI+ digital economy as a core industry track has met two conditions, that is, the current new energy track has entered a downward period, and at the same time, the policy continues to promote AI+ digital economy, and only the development factors are still relatively lacking. Compared with the development of smartphones: (1) the iPhone was released in January 2007 and officially released in June; In 2008, HTC released the first Google Android smartphone HTC Dream, and began to gradually enter the smartphone era. (2) The iPhone 4 was released in April 2010 and the iPhone 4s was released in October 2011 to accelerate the development of smartphones. (3) Domestically, Xiaomi was established on April 6, 2010, Xiaomi announced its entry into the market in July 2011, and Xiaomi mobile phone sales exceeded 1 million units in March 2012; In May 2011, Gionee established a smart phone R&D institute, launched a smart strategy and held a launch conference for all series of smart phone products in November; In June 2011, OPPO launched its first smartphone, the OPPO X903 (Find) and more. Compare the development of the AI+ industry: (1) On June 11, 2018, OpenAI released the GPT-1 model; (2) GPT3.5 was released in November 2022, followed by the release of the large-scale multimodal model GPT-4 on March 15, 2023; (3) Baidu and iFLYTEK launched the Wenxin model and iFLYTEK Spark model respectively, and Tencent, Huawei, Ali, etc. also accelerated their layout. Therefore, if you consider the entry of domestic giants and the R&D cycle, from the emergence of overseas explosive models to the explosive growth in China, the time cycle can be achieved in about 1-2 years; Analogous to the current leading enterprises began to accelerate the layout of large models, it is expected to be implemented in a cycle of about 1 year.

The reason why the metaverse cannot become an industry-level market is that it is difficult to implement due to technical difficulty and lack of ecology. Compare the difference in the theme market of the metaverse: First of all, from the performance point of view, the performance of the metaverse has grown rapidly in 2021Q1-2021Q4, but the performance growth rate has continued to decline since 2022, so the weak downward performance of the metaverse drags down the market performance, resulting in the market being unsustainable. Secondly, from the perspective of industrialization conditions: (1) the difference in technical foundation and the higher threshold of metaverse technology: on the one hand, technologies such as AI-based digital economy have already accumulated to a certain extent, including speech recognition, image recognition, etc. have been widely commercialized; On the other hand, the metaverse has a higher technical threshold (cloud rendering, VR anti-stun, real-time interaction, etc.), in contrast, the increasingly large-scale AI models and massive data can accelerate the development of AI technology and have a broad landing foundation. (2) Application scenarios and ecology are relatively scarce: Up to now, only Internet giants such as Google, Meta and Microsoft have released related products, and Oculus Quest 2 has not been followed by a large number of manufacturers after its release; At present, the content side of the metaverse is relatively single, mainly in the field of games and entertainment, and the ecological threshold is relatively high, so it is difficult to widely commercialize it in the short term. In contrast, digital technology, led by AI, has a wealth of potential application scenarios, including travel, social networking, and office. (3) Differences in policy support: On the one hand, policies related to data elements are frequently introduced, mainly for the promotion of property rights, circulation and trading systems of data elements, and vigorously develop infrastructure construction, including computing power infrastructure based on "counting data from east to west"; On the other hand, the national top-level design, the establishment of the National Data Bureau, the AI Promotion Office, etc.; In contrast, in the metaverse, only some provinces have released relevant industrial development plans.

Special Report on Artificial Intelligence Industry: The Apocalypse of Historical Experience on the β of AI Industry

2. The β market in the upward trend of the industry

2.1. Mobile Internet 2012-2016

The core drivers of the Internet market are profitability and liquidity. Review of the market performance of the development of the mobile Internet industry from 2012 to 2016: (1) Market performance is not equivalent to industrial development, if from the perspective of excess income, the market began in December 2012, that is, the profits of industrial enterprises bottomed out and rebounded, ending in July 2016. (2) Compared with profits, it can be seen that the beginning and end of market performance basically coincide with the upward trend and peak of profits, that is, when TMT profits bottom out and the profit growth rate is greater than that of All A, the market starts, and the TMT profit growth rate peaks and falls back and the market ends when All A profits bottom out; However, TMT's earnings from early 2014 to June 2015 were relatively weak, but the reason for the obvious excess returns was the liquidity-driven "leveraged bull" market. (3) Comparing the valuation performance, it can be seen that the mobile Internet market can be roughly divided into three segments, the first paragraph 2012/12-2013/12 is the earnings valuation resonance under Davis double-click, the second paragraph 2014/1-2015/12 is the liquidity driven by the valuation market under weak profits, and the third paragraph 2016/1 and later boosts market performance under profits under tightening liquidity.

The mobile Internet market is divided into four stages based on smartphone penetration rate and the number of mobile Internet users. In order to further discuss the rhythm of sub-sectors in TMT, we divide different industrial stages by smartphone penetration rate and mobile Internet user growth rate, and combine the market performance as follows: (1) the initial stage of the industry (2012/12-2013/10): Under the background of the bottoming recovery of industrial enterprise profits in December 2012, the performance of the A-share market began to strengthen, and the smartphone penetration rate continued to rise to the central level of 72%; (2) Industrial development period (2013/11-2014/12): the smartphone penetration rate has further increased to the center of 87%, and the growth rate of mobile Internet users has continued to rise to 8.38% under the background of continuous economic repair, and the number of users has reached 875 million; (3) Industry outbreak period (2015/1-2016/2): the smartphone penetration rate rose slightly but reached a bottleneck, while the growth rate of mobile Internet users further rose to 14.08%, and the number of users reached 1.019 billion; (4) At the end of the industry (2016/3-2016/12): the growth rate of mobile Internet users remained at 11%-16%, and the number of mobile Internet users reached 1.094 billion in December 2016; After that, the dominant market of TMT ended, so the discussion will not continue.

In the early stage of the industry (2012/12-2013/10), looking at the performance and valuation flexibility, media> computers> communications, > electronics. Judging from the market performance in the first stage, after a period of general rise, the media > computers> communications, > electronics. (1) In terms of performance, the media > communications, electronics, > computers. Compared with the performance of 2012Q1 to 2014Q1, the media industry performance is stable and continues to grow, from the low point of -0.41% in 2012Q1; The performance of electronics and communications only bottomed out in 2012Q4, and there was some adjustment in 2013Q2; Computer profitability is relatively weak and less resilient. (2) In terms of valuation performance, the price-to-earnings ratio of computers and media is basically the same as that of electronics. In the first stage, the valuation of all segments of TMT increased significantly, among which the valuation of computer and media rose more, and the PE at the end of the stage was basically the same as that of electronics and communications. (3) On the whole, in the early stage of the industry, policies and industrial events continued to catalyze the general rise of TMT, among which the media had a stable and sustainable performance compared with electronics and communications, so it was obviously superior; Although the performance of computers is poor, because the stock market value and valuation are also lower and more flexible than other TMT industries, the media > computers> communications > electronics.

The industrial development period (2013/11-2014/12) looks at the penetration rate and performance changes, computer> electronics, communication > media. Compared with the market performance of the second stage, computers > electronics, communication > media, but their resilience is weaker than that of the first stage. (1) In terms of performance, electronics, communications and media all fell at a high level, but the absolute growth rate of communication was higher than that of electronics and media; Although the performance growth rate of computers was weak, it rebounded slightly. (2) The continuous catalyst of industrial development has brought about the β market of the TMT industry, and the penetration rate of different sub-sectors has continued to increase, especially upstream hardware, namely base stations, chips and consumer electronics; Among them, the 3G penetration rate was 33.94% at the beginning of 2014, slowly rising to a peak of 37.73% at the end of 2014, and the 4G penetration rate also began to rise sharply at this time. The growth rate of mobile phone shipments has continued to decline since 2012, and bottomed out in mid-2014; Online games and movie screens were in a downward cycle in 2014; The computer industry has benefited from the increase in upstream hardware penetration, and the number of employment and enterprises has continued to rise. (3) Generally speaking, in this stage, the computer industry penetration rate continued to rise and the performance picked up slightly, and the performance of computers was the strongest, and the performance of electronics, communications, and media was relatively weak under the downward performance and slowing down in penetration, so computers > electronics, communications, > media.

Special Report on Artificial Intelligence Industry: The Apocalypse of Historical Experience on the β of AI Industry

The industrial explosion period (2015/1-2016/2) looks at the flexibility of application-side performance, computer> communication, media > electronics. From the perspective of the third stage of market performance, all industries have risen sharply, among which computers> communications, media > electronics. (1) In terms of performance, computer earnings rose in 2015, while media and communications bottomed out in 2015Q1, and electronics profits declined; At the same time, the performance of computers in 2016Q1 rose sharply, much higher than other industries. (2) In the stage of industry outbreak, midstream software and hardware, and downstream applications entered a blowout period, and the activity of content, mainly games and videos, continued to increase, and the number of mobile Internet users rose from 4.62% to 14.08% year-on-year. (3) Micro liquidity abundant to promote the "leveraged cattle" market. Technology-based growth companies represented by TMT and others were obviously driven by loose liquidity and increased risk appetite by policies, but then the China Securities Regulatory Commission issued the "Opinions on Cleaning Up and Rectifying Illegal Securities Business Activities", and the micro-liquidity was greatly adjusted against the background of tightening. (4) Overall, the middle and downstream application ends at this stage have entered a blowout period to promote the TMT boom, loose liquidity boosts the industrial market, and the performance elasticity of the application side is expected to determine the market performance> computer communications, media > electronics.

At the end of the industry (2016/3-2016/12), return to value, electronics, communication> computer > media. Judging from the market performance in the fourth stage, TMT performed relatively weakly, among which electronics, communications> computers, > media. (1) In terms of performance, electronic profit bottomed out at 1.2% in 2016Q2 and continued to grow to 61% in 2017Q2; Communications also recovered slightly from the bottom in 2016Q1; Computers and media peaked in Q1 2015 and retreated, and then benefited from economic recovery Media earnings picked up. (2) From the perspective of the industry, communication base stations benefit from 4G construction and enter a new round of upward cycle; Semiconductors enter an upward cycle driven by global demand. (3) In terms of investment style, the investment preference of foreign capital after the opening of the Shanghai Stock Connect in 2014 continued to affect the style of the domestic industry, and after the opening of the Shenzhen Stock Connect in 2016, foreign capital inflowed sharply, and the core assets represented by the Shanghai and Shenzhen 300 were favored by foreign capital, and the investment style was more performance-driven. (4) On the whole, the bottleneck of the penetration rate at the end of the industry and the decline in profits, the overall performance of TMT is relatively weak, and the investment style returned to performance-driven under the large inflow of foreign capital, so the electronics, communications> computers > media with better performance.

The main reasons for the correction are tightening liquidity and external shocks, which often last 20 days. Looking back at the adjustment stage of the TMT market with a decline of more than 10%, there were a total of 6 times, specifically: (1) affected by the stock market crash or circuit breaker, the decline was often deeper, more than 30%, while the other 3 times were about 13%; (2) the remaining average number of days, excluding the adjustment in February 2014, was 20 days; The longer adjustment period in February 2014 was mainly due to weak macroeconomic fundamentals and lack of obvious catalysis. (3) The main reasons for the adjustment are: one is the tightening of liquidity, including the Fed's release of withdrawal QE signals, cleaning up over-the-counter capital allocation, etc.; the second is exchange rate risk and overseas market fluctuations; The main reasons for the termination of the adjustment are: one is to provide liquidity support, including the suspension of IPOs, the extension of the two financing facilities, etc.; The second is that the fundamental data is stable and improving; The third is the subsidence of negative factors in the previous period, including exchange rate stabilization and economic repair.

2.2. New Energy 2018-2022

The core of the new energy market lies in the performance-driven Davis double-click. Review of the market performance of the new energy industry development from 2018 to 2022: (1) from the perspective of excess income, the start time of the market is October 2018, that is, the time when medium and long-term loans bottom out; Ends July 2022. (2) Comparing the profit performance, it can be seen that the excess income performance of the new energy sector and the performance growth rate difference of new energy relative to all A are basically the same, from October 2018 to March 2020, there is basically no excess return under the weak performance advantage of new energy relative to all A, and the continuous high growth of new energy performance after March 2020 has led to new energy to obtain obvious excess returns, of which short-term adjustment occurs when the profit in Q1-Q2 of 2021 declines. (3) Compared with the valuation performance, it can be seen that the increase in valuation also tends to follow the earnings performance, so the core of the overall market promotion of new energy lies in the Davis double-click driven by earnings; The obvious downward trend in the market after 2022 is mainly due to the fact that the performance has not obviously declined.

The pricing determinant of the new energy segment is performance. Compared with the four main sub-sectors of new energy, (1) the market performance of lithium batteries, new energy vehicles and photovoltaics is more consistent with the performance trend, but among them, the wind power sector is often more flexible due to the small number of individual stocks and relatively low market value, so the performance elasticity is often large and deviates from the market performance. (2) Due to the large differences in the profit models and correlation of different sectors, comparing the performance of different industries can be used as a key indicator to track performance: lithium batteries mainly drive prices upward due to downstream demand, new energy vehicles increase sales under the promotion of policies, wind power tracks the growth rate of new equipment capacity, and photovoltaic tracks production capacity and prices.

Special Report on Artificial Intelligence Industry: The Apocalypse of Historical Experience on the β of AI Industry

The strength of performance determines the flexibility of the new energy segment. In September 2020, China clearly put forward the goals of "carbon peaking" in 2030 and "carbon neutrality" in 2060: (1) In the first phase (2018Q4-2020Q3), new energy is mainly driven by independent regulation by the market, and the profit growth rate of photovoltaic and wind power is positive while the growth rate of lithium batteries and new energy vehicles is negative, so it is manifested as photovoltaic, wind power> lithium battery, and new energy vehicles; (2) In the second stage (2020Q4-2022Q1), new energy enters the fast lane driven by the "dual carbon" goal, and photovoltaic profits are stable and continue to grow, and the growth space is certain, so the performance is the strongest; The growth rate of lithium batteries and new energy vehicles has bottomed out and is stronger than that of wind power, so it shows that lithium batteries, new energy vehicles > wind power; The sales growth rate of new energy vehicles continued to increase high due to the automobile consumption policy, and then drove the demand for upstream lithium batteries, and the price of lithium carbonate then began to rise sharply, so the rhythm of new energy vehicles → lithium batteries; (3) In the third stage (2022Q2-present), except for the performance of photovoltaics, the performance of the new energy sector began to decline, and the market has adjusted significantly.

The main reasons for the adjustment were external shocks and policy tightening, and the depth of the decline was about -20%. Looking back at the adjustment stage of the new energy market with a decline of more than 10%, there were 5 occurrences, specifically: (1) the depth range of decline was 16%-27%; (2) Adjustment under the influence of external shocks, the adjustment time lasts about 30 days; Adjustments affected by economic cycles take a long period of time. (3) The main reasons for ending the adjustment are: first, the marginal easing of liquidity; The second is fundamental catalysis.

2.3. Implications for the AI+Digital Economy

Summary of the review conclusions of mobile Internet and new energy in 2012-2016 and new energy in 2018-2022: The β market began at a major economic inflection point. The starting point of the A-share market is not the same as the starting point of industrial development, and the starting point of the β market often begins at a major inflection point in the macroeconomy, such as the mobile Internet begins with the upward trend of industrial enterprise profits, the new energy source begins with the upward trend of medium and long-term loans, and the current round of AI+ digital economy begins with the optimization of epidemic prevention and control. Performance is the core of determining β market. Comparing the performance performance, we can see that (1) the trend of performance is the core of the continuation of the industrial market, that is, the upward growth rate of performance brings the upward trend, and if the performance declines in the industrial market, there will be a short-term adjustment. (2) In the upward market of the industry, if there are other positive variables such as sufficient micro funds, it can promote a sharp upward trend in the market, such as the "leveraged bull" in 2014 and the high increase in new funds in 2020. (3) The divergence of the performance of the sub-sector is mainly affected by the performance elasticity, that is, the performance with higher performance growth rate / higher expectation is relatively good. At present, computer and media profits are rising, so although computer and media have adjusted in the short term, the upward trend remains unchanged in the long run; While the profits of communications and electronics have declined, communications mainly benefited from the influence of the special estimate The three major operators with higher weights rose more, and although electronics rebounded significantly in the short term, from the long-term trend, it is still in the downward range.

The spread of market conditions in the β is mainly determined by the rhythm (penetration rate) of the industry itself. (1) The performance of the mobile Internet market is: media→ computers→ communications, electronics, compared with the penetration rate, we can see that we believe that the main thing is that the performance of media (especially games, movies, etc.) in the initial stage of the industry is the easiest to cash, the market gradually turns from the blue ocean to the red ocean development brings a high demand for software and hardware in the middle and upper streams, and finally the supply and demand ends under the gradual saturation of the return to value, and the industrial market ends. (2) The performance of the new energy market is: photovoltaic, wind power→ photovoltaic, new energy vehicles→ photovoltaic, lithium battery, compared with industrial development, under the early promotion of energy structure transformation, photovoltaic and photovoltaic will enter the upward cycle, driven by technological upgrading to bring about the economic improvement of new energy vehicles, the penetration rate continues to increase under the new energy vehicle performance is better, and then the demand for new energy vehicles increases to drive the development of lithium batteries. (3) At present, we believe that the media that is the easiest to cash in on the performance of economic repair and regulatory relaxation will take the lead in benefiting from the software and hardware equipment of the midstream that → promote industrial expansion, including data, computing power infrastructure, etc. → large models gradually landed to promote midstream software and downstream applications under the outbreak of downstream application scenarios.

Special Report on Artificial Intelligence Industry: The Apocalypse of Historical Experience on the β of AI Industry

β the elasticity of the market is high-low-high, and the adjustment is mainly affected by external shocks, policy or liquidity tightening. (1) From the perspective of the two industrial markets, after experiencing a short-term rapid upward trend, it will experience a long period of shock upswing, and the overall elasticity is obviously weak, such as 2013/12-2015/1 in the mobile Internet market (TMT index +57.63%, Wande Quan A+59.94%), 2019/5-2020/11 in the new energy market (new energy +0.28%, Wande Quan A+1.19%). (2) The adjustment of industrial conditions is mainly affected by three aspects: one is external shocks, such as epidemic shocks, exchange rate risks, etc., the second is the tightening of policies or liquidity, such as cleaning up over-the-counter capital allocation, etc., and the third is the weakening of the macro environment, such as the decline in profits of industrial enterprises. (3) The recovery of the industrial market is catalyzed by three aspects: one is the loosening of liquidity, such as interest rate cuts and RRR cuts, the second is the repair of fundamentals, such as the rise in corporate profits, and the third is the subsidence of the previous suppressive factors, such as the stabilization of the exchange rate. Compared with the current point of view, (1) has experienced a rapid upward stage, or will enter a long period of shock upward, mainly we believe that the current economic repair momentum is weak under the decline in market sentiment, and the large model and other landing tasks are heavy and long, so we need to see more positive signals to continue to promote the continuation of the market. (2) Compared with the adjustment range and span, the adjustment range is 20-140 days, and the adjustment range is -11%--45%; Therefore, the current short-term adjustment is still within a reasonable range.

3. Look for super α in the β

3.1. Features of α

The super α in the β market is gradually shifting from market value driven to performance driven. Comparing the size of the market capitalization, the growth rate of the performance and the change ratio of the market value in the two markets, it can be seen that: (1) the super α in the mobile Internet market from 2012 to 2016 tend to be companies with a small market value, and the market value is concentrated between 5-10 billion; From 2018 to 2022, the correlation between super α and market value in the new energy market is weak, and the market value is concentrated between 2 billion and 20 billion, and the market value has increased. (2) In the mobile Internet market from 2012 to 2016, it did not show that the better the performance, the greater the flexibility; In contrast, in the new energy market from 2018 to 2022, the better performance tends to be more resilient to growth. Therefore, the super α in the two β markets has gradually shifted from market value driven to performance driven, we believe: first, the deepening of institutionalization, market pricing is increasingly focusing on profit performance; Second, the structural market gradually replaces the general rise market, so it presents a situation in which the strong are always strong, and the leading effect is significant; Third, the shell value weakened, and the speculation of small market value gradually dissipated.

3.2. Review and tracking of super α

The fund's heavy stocks serve as a pool of super-α stocks. Due to the difference in market structure and pricing, we choose fund heavy stocks as the stock pool tracked by super α, mainly because: (1) the long cycle in the industrial market, the number of listed companies involved, and fund heavy stocks often have more investment value, including long-term allocation value, growth or theme speculation;α (2) It can save the time of the fund company's shares in the pool or have met the requirements for entering the pool.

Super α tracking is based on the number of analyst reports on the company. We reviewed the new energy market from 2018 to 2022 to find super α, and selected 14 tracking stocks from the heavy stocks disclosed by the fund in 2017; (1) The market performance of super-α is driven by performance, but the disclosure of performance is often lagging, and it is necessary to track high-frequency microdata to judge the performance trend; (2) High-frequency data is effective but often requires greater effort to identify information, so future performance can be judged through analysts' consensus forecasts for future profits; (3) Predicted market value FY1 and the actual trend convergence, and the predicted market value FY1 / actual market value is greatly affected by market fluctuations, and the ratio is > 1.5 when the market is rising or falling rapidly; (4) Therefore, we use the number of reports as a key indicator to predict market trends: on the one hand, the number of reports can indicate the market's optimism about the company, and on the other hand, the report will gradually revise the company's profit expectations, so increase or reduce positions.

The number of LONGi green energy reports was roughly ahead of the market price in the first quarter: from 2017Q3 to 2020Q2, the increase in the number of reports in the previous quarter was often accompanied by the upward trend of the stock price in the following quarter; Stock price adjustment phases, such as 2018Q2, Q3 and 2020Q1, are also expected to decrease in the number of reported figures in the previous quarter. In the post-2021 stock price decline phase, the number of reports began to decline all the way from the fourth quarter of 2020.

The number of Sungrow reports changes synchronously with market prices: from 2017 to 2020, market prices and reported numbers oscillate synchronously; From the second quarter of 2020 to the end of 2021, market prices rose along with the increase in the number of reports; The decline in share prices in the first and fourth quarters of 2021, the first quarter of 2022, and the decline in the first quarter of 2023 also corresponded to the decline in the number of reported prices.

The number of super α reports is basically ahead of the curve or synchronizing the market performance. Of the 14 stocks selected, 3 stocks had a reported number ahead of the change in stock price, 10 stocks reported a change in the number of reports that were in sync with the stock price, and only 1 stock had a trend that did not match, with a 92.9% win rate.

Reduced or no reporting is more pessimistic about the corresponding stocks. (1) There was no company-related report on cisna shares in 2017-2019, and the corresponding stock price also fell by nearly 3/4 in early 2018, at this time, although the fund's position was higher, but with the collapse of the stock price, the proportion of holding decreased rapidly; In 2019-2020, although the number of reports occasionally increased, the number of reports per month was less than 2, and the stock price and fund holdings did not show a significant upward trend. (2) Senyuan Electric has company-related reports from 2017 to 2019, but the number of monthly reports is small, and the stock price has fallen slightly; After 2019, the number of company reports plummeted to zero, and the share price and fund holdings began to decline rapidly.

Special Report on Artificial Intelligence Industry: The Apocalypse of Historical Experience on the β of AI Industry

In summary, we can use the fund's heavy stocks disclosed each quarter as a stock pool, and then increase or decrease positions according to the change in the number of reports. In the current AI+ industry market, the corresponding sample is selected to compare the changes in the number of reports to increase or reduce positions: (1) The stock prices of Newpoint Software and Sanwang Communication are synchronized with the number of reports, of which the number of reports of Newpoint Software has rebounded, but the market performance continues to decline, so you can choose to increase positions, and the number of reports and stock prices of Sanwang Communication are in a downward trend, so there is still a risk of continuous adjustment in the short term; (2) The trend of changes in the number of reports reported by Zhiyuan Internet and Mango Super Media is about one quarter ahead of the stock price change; Therefore, the current Zhiyuan Internet and Mango Super Media report can be increased under the background of the recovery of the number.

(This article is for informational purposes only and does not represent any investment advice from us.) For information, please refer to the original report. )

Selected report source: [Future Think Tank]. 「Link」

Read on