Source: CICC Insights
Authors: Liu Gang, Yang Xuanting, Zhang Dian
Over the past year or two, without the unexpected rise of AI, global growth might have faced even greater pressure amid weak traditional demand. For example, the United States' investment of up to one trillion US dollars in technology hardware and software contributed one-third of the GDP growth expected by 2025 (Chart 1), not to mention the potential boost to future growth from the increase in factor productivity (Chart 2).
AI's contribution to the stock market has also been significant. Since the release of ChatGPT at the end of 2022, the Mag7 stocks in the US market have contributed$S&P 500 Index (.SPX.US)$45ppt out of an 84% return, accounting for more than half (Chart 3). Since the release of DeepSeek in early 2025, the seven leading tech stocks in Hong Kong once contributed$Hang Seng Index (800000.HK)$14ppt out of a 37% return, accounting for 40% (Chart 4). Not only in China and the US, but in 2025, global market leaders such as South Korea, Japan, and Taiwan are also key links in the AI industry chain (Chart 5).
Chart 1: Among the average annualized quarter-on-quarter real GDP growth of 2.5% in the first three quarters of 2025, U.S. tech hardware and software contributed 0.8ppt.
Chart 2: Since 2023, labor productivity in the U.S. non-farm business sector has increased by 7.2%.
Chart 3: The Mag7 stocks in the U.S. market account for 45ppt out of the S&P 500's 84% return, making up more than half of it.
Chart 4: Seven leading tech stocks in Hong Kong account for 14ppt out of the Hang Seng Index’s 37% return, making up 40%.
Chart 5: AI-driven style leads again in global markets since the beginning of the year.
However, alongside the enthusiasm for AI is an ever-present concern about bubbles. After three years of rapid development, very few people now question the fundamental prospects of the AI industry itself. However, concerns remain regarding the possible gap between the ultimate realization speed and the level of investment. Just like the dot-com bubble of the 1990s laid a solid foundation for the booming mobile internet in the 21st century, it did not prevent the bursting of the bubble in 2000 from squeezing out excessive investment. Therefore, the bubble itself is not necessarily a bad thing, as it also drives industrial development. Discussing whether it will turn into a bubble is not particularly meaningful; what is more important is to confirm the current stage of development.
In this process, investment plays a crucial role as the source and destination of funds directly determine their behavior and investment orientation. As the 'two poles' of the global AI industry landscape, China and the United States exhibit significant differences in funding sources and investment directions due to disparities in areas such as computing infrastructure, chips, and models. Tracing these differences can help us understand variations in development paths and provide insights into different investment directions.
Chart: Both China and the U.S. allocate approximately 87-88% of investments to the foundational layer, with the technical layer accounting for 12-13%.
Chart: The foundational layers of China and the U.S. show industrial chain interconnections; chips exhibit autonomous fragmentation, while the application layer demonstrates mapping.
AI Industry Landscape in China and the U.S.: The U.S. has a first-mover advantage, while China is rapidly catching up. The U.S. faces a 'power shortage,' China a 'chip shortage,' with limited differences in models.
The cornerstone of the artificial intelligence industry lies in computing infrastructure, models, talent reserves, and financial support from capital markets. In the early stages of development, the U.S. held considerable leading advantages in computing infrastructure, models, advanced talent development, and data quality. However, following the release of DeepSeek in early 2025, China achieved breakthrough progress in the model layer, particularly in the performance of open-source models (Chart 6), and began narrowing the gap with the U.S. across multiple fields.
Chart 6: China achieves breakthrough progress in the model layer, particularly in the performance of open-source models.
► Computing Power Infrastructure: This encompasses physical infrastructure centered on data centers, digital infrastructure represented by chip R&D, and the mobilization and delivery of computing power via cloud computing. In the 2025 'Cloud Computing Blue Paper' published by the China Academy of Information and Communications Technology (CAICT), Gartner's data was cited showing that the global cloud computing market size reached $692.9 billion in 2024, with North America dominating 54.3% of the market share, while China accounted for 16.8%. By 2025, China’s share is expected to rise further to 18.3% (Chart 7).
Chart 7: Global cloud computing market size in 2024 reached $692.9 billion, with North America accounting for 54.3% and China for 16.8%.
1) The United States has a first-mover advantage in infrastructure, but China is narrowing the gap underpinned by its superior power supply. As a result, the U.S. is more 'power-constrained'.
Although the United States currently far surpasses China in terms of server and data center numbers (Chart 8), China has a higher density of computing power capacity. Taking data centers as an example, the number of data centers in the U.S. is more than eight times that of China (4,165 vs. 500), but their capacity is only 1.7 times greater (53.7GW in the U.S. vs. 31.9GW in China). More importantly, large-scale deployment of data centers requires substantial electricity support. By 2024, China's electricity generation had already exceeded twice that of the United States (Chart 9). The electricity consumption of existing data centers in the U.S. accounts for 4.4% of its total electricity usage, whereas in China it is only 1.1%. In an executive order signed by Trump in January 2025, it was explicitly stated that 'new large-scale AI infrastructure projects must be accompanied by the construction of additional clean energy power generation facilities to avoid encroaching on residential electricity demand.'
Chart 8: The U.S. currently far surpasses China in the number of servers, physical data centers, and cloud computing facilities.
Chart 9: By 2024, China’s electricity generation had exceeded twice that of the U.S.
2) The U.S. dominates chip R&D, while China accelerates its localization efforts. However, there remains a gap in advanced process technologies.
According to statistics from the Semiconductor Industry Association (SIA), sales in the U.S. semiconductor industry reached USD 318 billion in 2024, accounting for 50.4% of global revenue. The sales shares of mainland China and Taiwan were 4.5% and 6.5%, respectively (Chart 10). Although these shares remain significantly lower than those of the U.S., China’s chip industry has been growing rapidly. In 2024, the total shipments of AI chips exceeded 2.7 million units, with local chip brands shipping over 820,000 units, marking a year-on-year increase of 310%. However, beyond achieving cost-effective alternatives in mature processes, continuous breakthroughs in advanced process technologies are even more critical.
Chart 10: In 2024, sales in the U.S. semiconductor industry reached $318 billion, accounting for 50.4% of global revenue.
► Model: The United States still leads in both overall quantity and quality, but China has gained an advantage in the open-source model field.
According to statistics from Epoch AI, among the 976 well-known models currently available, the number of models in the U.S. is four times that of China (632 vs. 156). However, in the Artificial Analysis Intelligence Index scoring, Chinese large models such as Zhipu GLM, DeepSeek, and Kimi closely follow behind major U.S. models like ChatGPT, Claude, and Google Gemini, ranking within the top ten (Chart 11).
China also leads in downloads and usage in the open-source model domain. According to statistics from the Atom Project, the cumulative downloads of Chinese open-source models surpassed those of the U.S. in August 2025 (Chart 12), and as of December 2025, over 62% of model derivatives were based on Chinese large models (Chart 13). To a certain extent, China has leveraged its model advantages to compensate for shortcomings in chips.
Chart 11: Comparison of ratings of major domestic and international models
Chart 12: Cumulative downloads of Chinese open-source models exceeded those of the U.S. in August 2025
Chart 13: As of December 2025, over 62% of model derivatives were based on Chinese large models, surpassing the 32% from the U.S.
► Talent Pool: China's attractiveness to top talents is gradually increasing, with patents surpassing those of the U.S.
The core driver behind the continuous advancement of AI technology lies in talent. According to the 'Global Artificial Intelligence Research Trends Report (2015-2024)' released at the Global Digital Economy Conference in July 2025, the number of AI researchers in the U.S. and China accounted for 57.7% of the global total. The U.S. leads globally with over 63,000 researchers, while the number of researchers in China increased from less than 10,000 in 2015 to 52,000 in 2024. The rapid growth in talent has strengthened China’s research capabilities in AI. By 2022, the number of AI patents filed by Chinese researchers had reached three times that of the U.S. (Chart 14).
Chart 14: By 2022, the number of AI patents held by Chinese researchers had reached three times that of the United States.
Overall, the United States has an earlier start in the field of artificial intelligence and holds a first-mover advantage in computing power and model development, giving rise to leading companies on a global scale. In contrast, China, benefiting from policy support, a large domestic market, and a talent dividend, is gradually narrowing the gap with the United States in the field of artificial intelligence. While the bottleneck for the United States lies in more fundamental infrastructure such as electricity, China faces challenges in the development of advanced process chips. The limited availability of open-source models further underscores the basic pattern that underpins the 'differences' in investment orientations between China and the United States.
China-US AI Investment Landscape: Similar investment intensity between China and the US; China shows stronger performance when considering infrastructure; greater macroeconomic impact on the US.
► Technology equipment investment: China’s nominal investment scale is 60% of that of the United States, but its share of GDP is comparable to that of the United States, both at 3.3~3.4%.
If we define narrow AI investment as technology hardware plus software equipment under GDP terms, in 2025, the United States will reach approximately $1.05 trillion, accounting for 3.4% of nominal GDP, which has increased by 0.5 percentage points since 2023. China's annualized scale of technology hardware plus software under the same criteria is approximately $650 billion (4.6 trillion yuan), equivalent to 60% of the United States, accounting for 3.3% of nominal GDP, similar to that of the United States (Chart 15).
Chart 15: In 2025, the proportion of US technology investment in nominal GDP will be 3.4%, close to China’s 3.3%.
► Investment including infrastructure: China’s share of GDP approaches 6%, higher than the United States’ 4.6%.
The AI industry chain is not limited to technology equipment. Considering the construction of data centers and power facilities, as well as R&D investments in related industries, in 2025, AI spillover demand in the United States may bring about an additional $400 billion in investment, raising the broad-scale AI investment to $1.4 trillion, accounting for 4.6% of nominal GDP (Chart 16). However, due to inconsistencies in GDP sub-item calculations between China and the United States, we use the increase in computing power to estimate the positive impact of AI investment on GDP.
Chart 16: AI-related investments in the U.S. may rise to $1.4 trillion by 2025, accounting for 4.6% of nominal GDP.
The China Academy of Information and Communications Technology (CAICT) noted in its 'Research Report on Computing Power Economy Development (2025)' that 'empirical analysis shows that every 1% increase in computing power scale will correspondingly drive a 0.425‰ increase in China's GDP.' According to IDC data, China’s computing power scale is expected to increase by 43% year-on-year in 2025, contributing an additional RMB 2.5 trillion to GDP (1.8% of total nominal GDP). The broad-based AI investment scale in China could rise to 5-6% (Chart 17).
Chart 17: A 43% increase in China's computing power scale in 2025 corresponds to an estimated rise in AI investment to 5-6%.
Economic impact: The U.S. information technology sector contributes three-tenths to GDP growth, while China's contributes one-tenth.
In assessing the macroeconomic contributions of the AI industries in the U.S. and China, we use the gross value added (GVA) approach under the production method to avoid distortions from equipment imports and capital goods pricing. When comparing the contributions of the IT sectors to GDP, the U.S. contributed 0.6 percentage points (ppt) to the 1.6% real GDP growth in the first half of 2025, representing 34% of the total. Meanwhile, China’s IT industry contributed 0.55 ppt to the cumulative 5.2% year-on-year GDP growth through the first three quarters of 2025, with a contribution ratio of 10.6%, up slightly from 9.6% in 2023 (Chart 18).
Chart 18: The U.S. IT industry contributes three-tenths of overall growth, while China's IT industry contributes one-tenth.
Differences in funding sources: The U.S. is predominantly driven by the private sector, whereas China is propelled by both government and private sectors.
Although the U.S. and China have similar proportions of total investment, differences exist in development speed and focus areas such as AI infrastructure, chip R&D, and model applications. A key reason lies in the differing funding sources for AI investments between the two countries. Funding sources determine attributes and behaviors like return expectations, time horizons, and investment destinations. These sources are divided into private and public sectors, with the private sector further split into corporate self-funding and venture capital.
Overall, U.S. AI investment is primarily led by the private sector ($552 billion), with limited direct government investment ($11 billion). In contrast, China’s private sector investment ($90 billion) is only one-sixth of the U.S.’s, but government direct investment and guided funding are significantly larger ($75 billion). Specifically,
► Private Sector: The scale of U.S. investment is larger (USD 552 billion), nearly six times that of China (USD 90 billion).
1) At the leading company level, U.S. investment is nearly five times that of China. In the U.S. stock market, we measure the total investment across the entire industrial chain by examining capital expenditures directed toward infrastructure by hyperscale cloud service providers, as well as R&D expenses for chips and large models (details below). By 2025, this figure has exceeded USD 400 billion. In the Chinese market, our estimate of the overall scale is approximately USD 84 billion (Chart 19).
2) At the venture capital level, the U.S. scale is 25 times that of China. According to PitchBook statistics, venture capital investment in China's AI sector in 2025 reached USD 6 billion, while AI-related venture capital financing in the U.S. amounted to USD 175 billion. Even after excluding expenditures related to model-layer investments by companies such as OpenAI and Anthropic to avoid double counting, the financing still reached a high of USD 152 billion (Chart 20).
Chart 19: By 2025, investment in foundational and technical layers by leading U.S. companies approaches five times that of China.
Chart 20: Scale of venture capital investment in AI sectors in the U.S. and China.
► Government Funding: China’s investment scale is larger (USD 75 billion), approximately seven times that of the U.S. (USD 11 billion).
The direct funding from the US government is much weaker compared to its private sector. The US government's budget for AI technology R&D increased from $8.2 billion in the fiscal year 2021 to $11 billion in the fiscal year 2025. Although the Stargate project was officially announced by Trump in January 2025, the core funding came from OpenAI and SoftBank, with some budgets overlapping with$Oracle (ORCL.US)$、Microsoft (MSFT.US)capital expenditures, and thus is not included in government investment.
The direct investment scale of the Chinese government may exceed CNY 500 billion (USD 75 billion). A major national-level direct investment fund in the AI field is the third phase of the National Integrated Circuit Industry Investment Fund (CNY 344 billion), with the National Artificial Intelligence Industry Investment Fund (CNY 60.06 billion) serving as its specialized sub-fund. Additionally, the Ministry of Finance led the establishment of the National Venture Capital Guidance Fund with an investment of CNY 100 billion. Together with capital expenditures from the three major telecom operators, the government investment is roughly estimated at over CNY 500 billion (USD 75 billion).
Differences in Investment Allocation: The U.S. invests more in data centers and supporting infrastructure, while China allocates more resources to chips and models.
From the perspective of investment direction, the AI industry can be divided into three main segments: 1) The foundational layer includes hardware computing power, covering core hardware such as AI chips, servers, and optical modules, as well as data center energy and supporting infrastructure, including liquid cooling equipment and electrical systems; 2) The technical layer focuses on innovation in large models and algorithm frameworks; 3) The application layer serves as the platform for technological implementation, encompassing vertical solutions across various industries.
► First, from the perspective of private sector investment: 1) The foundational layer focuses on capital expenditures directed towards infrastructure by Chinese and American cloud providers (US:$Amazon(AMZN.US)$、 Microsoft (MSFT.US) 、 $Google-C (GOOG.US)$ 、 $Meta Platforms(META.US)$ 、$Oracle (4716.JP)$、$CoreWeave(CRWV.US)$; China:Baidu (BIDU.US)、$Alibaba(BABA.US)$、 $Tencent (00700.HK)$, cloud vendors such as ByteDance), as well as the expenses incurred by major chip manufacturers for chip research and development (United States:$NVIDIA(NVDA.US)$、 Advanced Micro Devices (AMD.US) 、Broadcom (AVGO.US)andQualcomm (QCOM.US); China: $Hygon Information Technology (688041.SH)$、 $寒武纪-U(688256.SH)$ 、 $Moore Threads (LIST23919.SH)$ 、 $Muxi Corporation-U (688802.SH)$ , Huawei, and Baidu and Alibaba's self-developed chips);
2) The technical layer focuses on R&D investments by leading large model companies (U.S.: OpenAI, Anthropic, xAI, and Google’s large models; China: Baidu, Alibaba, Tencent, ByteDance, Zhipu, Minimax, etc.); 3) The application layer spans various industries, making it difficult to achieve precise disaggregation and statistics.
Overall, excluding investments in the application layer, by 2025, the scale of investment by leading U.S. companies in the foundational and technical layers is expected to be five times that of their Chinese counterparts ($400 billion vs. $84 billion), with expectations that this gap will widen further by 2026.
1) U.S.: Of the $400 billion, 88% is allocated to the foundational layer (primarily data centers and supporting infrastructure at 83%, and chips as a secondary focus at 5%), while 12% is allocated to the technical layer (models). In 2025, the U.S. foundational layer investment is projected to reach approximately $3,500 billion, representing a 2.4-fold increase from 2022. Among this, data center and related infrastructure investments are estimated at $3,340 billion (83%), while chip R&D accounts for a relatively smaller portion of total investments (5%), remaining stable between $15-20 billion. Investment in the technical layer has grown rapidly since 2022, rising from $4.2 billion to $48 billion by 2025, increasing its share of total investment to 12% (Chart 21).
Chart 21: Ninety percent of the $400 billion invested by leading U.S. companies goes to the foundational layer, with ten percent going to the technical layer.
2) China: Of the $84 billion, 78% is allocated to the foundational layer (data centers and infrastructure at 70%, with a relatively larger share for chips at 8%), and 22% is allocated to the technical layer (models). By 2025, China's foundational layer investment is projected to reach approximately $65 billion, with data centers and supporting infrastructure accounting for $59 billion (70%) and chip R&D reaching $6.4 billion (7.5%). Investment in the technical layer is estimated at $19 billion (22%) (Chart 22).
Chart 22: Nearly eighty percent of the $84 billion invested by leading Chinese companies goes to the foundational layer, with twenty percent going to the technical layer.
► Secondly, looking at government investment: 1) U.S. government funding prioritizes basic research in the technical layer and cutting-edge directions in the application layer, with an $11 billion budget primarily directed toward non-commercial fundamental research in artificial intelligence, such as novel algorithms, and AI+ initiatives.
2) Chinese government funding focuses on chip R&D and 'hard technology' in the foundational layer, such as the third phase of the National Integrated Circuit Industry Investment Fund (344 billion yuan), which explicitly allocates 70% of funds to domestic equipment and materials, and 30% to advanced packaging and AI storage, emphasizing heavy asset industries like semiconductor manufacturing with long cycles, with a duration of 15 years; the National Venture Capital Guidance Fund has a duration of 20 years, making it the longest-standing 'patient capital' in China, adhering to the principles of 'investing early, investing small, investing long-term, and investing in hard technology'.
Aggregating both private and government investments, the proportion of foundational layer investments in the U.S. and China is around 87-88%, while the technical layer accounts for 12-13% (Chart 25). The U.S. foundational layer investment amounts to $3,500 billion (funded by leading companies), with technical layer investments totaling approximately $53.5 billion (comprising $48 billion from leading companies and 50% of $11 billion in government funding), representing shares of 87% and 13%, respectively (Chart 23); China's foundational layer investment totals $140 billion (including $65 billion from leading companies, $11.4 billion from the three major telecom operators, and $63.4 billion from government funds), with technical layer investments amounting to $19 billion (funded by leading companies), representing shares of 88% and 12%, respectively (Chart 24).
Chart 23: The overall investment proportions in the foundational and technical layers in the US are 87% and 13%, respectively.
Chart 24: The overall investment proportions in the foundational and technical layers in China are 88% and 12%, respectively.
Chart 25: The investment proportions in the foundational layer in both the US and China are approximately 87-88%, while the proportions for the technical layer are 12-13%.
Implications of Differences in AI Investment between the US and China: Private sector dominance in the US is driven by return constraints, focusing on infrastructure; China's government-led approach emphasizes chips and models.
Differences in industrial structure, funding sources, and investment allocation between the US and China directly shape their respective development trajectories and investment patterns. Several insights emerge:
1) AI investment in the US is predominantly led by the private sector, with its core driver being the pursuit of commercial returns. In the short term, the ability to mobilize resources for public infrastructure coordination is weaker compared to China. This is one reason why China has rapidly narrowed the gap in AI infrastructure under policy support in recent years. On the other hand, the characteristic of private capital dominance means that once returns fall short of expectations or are delayed, the market becomes highly susceptible to bubble concerns.
2) AI investment in China is government-led, employing “patient capital” for long-term strategic guidance. This model demonstrates exceptional resource allocation capabilities, enabling investments in critical areas such as computing power and chips without regard for short-term profitability. It exhibits strong investment resilience but faces challenges related to lower sensitivity to financial returns.
3) US AI investment focuses on data centers and energy infrastructure. According to Aerio’s statistics, there are currently 628 data centers under construction in the US. Investors, primarily hyperscale cloud providers, must ensure that substantial upfront investments are not stalled by power supply bottlenecks, which would otherwise lead to wasted capital expenditures and deteriorating financial performance.
4) Chinese AI investment prioritizes the foundational layer, particularly chip development. From the current state of the AI industry discussed earlier, a gap remains between China and the US in computational infrastructure, especially in advanced process technologies for chip development. Based on the current allocation of funds from enterprises and the government, data centers and bottleneck-prone sectors remain the primary investment focus.
How can the AI industries of the US and China collaborate? There is industrial chain synergy in the foundational layer, divergence in chip autonomy demands, and mutual learning and reflection in the application layer.
The high connectivity and portability inherent in the AI industry, combined with differences in resource endowments, sources of funding, and investment directions between China and the US, have resulted in industrial chain linkages at the foundational level of the AI sectors and markets in both countries (such as the value chains required for chip manufacturing and data center construction). However, geopolitical fragmentation has also driven demand for domestic chip substitution. At the application level, there is more reflection and mutual learning regarding commercial scenarios.
► The core of the foundational linkage lies in the joint pull of investments from both China and the US on the demand for related industrial chains. As analyzed above, the US faces a greater shortage of data centers with high computational density and power infrastructure. Investments in these areas will also drive demand for Chinese companies with competitive advantages in certain segments of the value chain, such as liquid cooling and power equipment accessories. Continued investment in chips (although not as significant as in China) will similarly boost demand for core hardware like optical modules and PCBs. The same logic applies to China, which also needs to continue investing in data centers, with an even larger proportion allocated to chips.
► The connection at the application level is mainly reflected in business model cross-referencing. Practical experiences and business models in similar vertical fields can provide references for each other, such as in the health AI sector (e.g., Ant Alipay’s Afo, ChatGPT Health, etc.). The recent heated discussion in the A-share market about the GEO concept is linked to expectations catalyzed by Elon Musk's announcement of plans to open-source the X platform recommendation algorithm soon. In the commercial exploration of AI assistants, the US has personal assistant apps like ChatGPT, while Apple has announced the deep integration of Google’s Gemini model into its ecosystem. Meanwhile, China has Qwen deeply integrated into Alibaba’s ecosystem, evolving towards becoming an AI assistant (Chart 26).
Chart 26: Industrial chain linkages in foundational layers between China and the US; chips exhibit autonomous fragmentation, while application layers show reflections.
From the perspective of the capital markets, 1) listed companies in the foundational layer are mainly concentrated in the A-share and US stock markets (such as NVIDIA, Broadcom, Qualcomm in the US stock market, Moore, Muxi, Cambricon, Hygon in the A-share market, and Wallachia in the Hong Kong stock market, as well as $New E-Sun (300502.SZ)$、$Zhongji Xuchuang (300308.SZ)$and$Tianfu Communication (300394.SZ)$and other optical module enterprises, including data centers, liquid cooling, energy storage, and power equipment, etc.), the market's expectations for revenue growth rates of listed companies in the foundational layer in China and the US are relatively higher than those in the technical and application layers (Chart 27, Chart 28), primarily due to more certain capital expenditures and policy-supported demand. In terms of market performance, excess returns of foundational layer companies in China and the US show a relatively high positive correlation but lack stability (Chart 29). This is partly because the foundational layer is more susceptible to trade frictions and partly due to the high volatility brought by high expectations and valuations;
2) Leading listed companies in the technology layer are more prevalent in the Hong Kong and US stock markets, including Hong Kong-listed companies such as Alibaba, Zhipu, Minimax, and Tencent, as well as US-listed companies like Google and Meta.
3) The distribution of companies in the application layer is relatively more balanced between A-shares and Hong Kong stocks, partly due to the broader range of vertical applications beyond internet platforms. From the perspective of excess return correlations, the correlation between China and the US has become more pronounced and stable after the launch of DeepSeek (Chart 30), reflecting the results of innovation linkages and business model cross-referencing.
Chart 27: Among Chinese assets, market expectations for profit growth rates in the foundational layer are relatively higher.
Chart 28: The US market is no exception, with higher expectations for the profit growth rate at the foundational layer.
Chart 29: Despite industrial linkages leading to a positive correlation in the performance of the foundational layers in China and the US, it is not particularly stable.
Note 1: Excess return calculations use Wind All A Index as the benchmark for A-shares, the S&P 500 for US stocks, and the Hang Seng Index for Hong Kong stocks, as shown on the right; Note 2: A-share PCB is represented by the CITIC PCB Industry Index, and optical modules are depicted by the Wind Optical Module Index. Source: Wind, CICC Research Department.
Chart 30: Following the advent of DeepSeek, assets in the technology and application layers of China and the US demonstrate a positive and stable excess return correlation.
Note 1: The A-share application layer is represented by the Guozheng AI Application Index, while the US stock application layer includes Adobe, Salesforce, and Palantir, represented by an equal-weighted average of the performance of major leading stocks.
Note 2: The US technology layer includes Google, Microsoft, and Amazon. The Hong Kong technology and application layer encompasses Tencent, Baidu, Alibaba, SenseTime, Kingdee International, Kuaishou, Meitu, Fourth Paradigm, Ali Health, and Kingsoft Software, represented by an equally weighted average of key leading stock performances. Source: Wind, CICC Research Department.
Looking ahead, the certainty of earnings realization in the foundational layer remains high, while the potential upside for the technology and application layers is greater. From the perspective of credit cycles, the technological chain represented by AI continues to be the primary growth driver. Specifically,
The US is still heavily investing in data centers and energy equipment, which will drive demand for supply chains in China's computing infrastructure (e.g., optical modules), data centers (e.g., liquid cooling), and energy infrastructure (e.g., related power and energy storage equipment).
China still requires chips, whether driven by capital expenditure in the foundational layer or by the national substitution strategy. Therefore, sectors such as semiconductors continue to have demand and policy support, but their downside lies in relatively high valuations and expectations.
The technology layer primarily focuses on the technical advancements of large models, with developments from both China and the U.S. potentially catalyzing each other.
On the application side, the catalyst comes from progress in vertical scenarios across various industries; if B2C business models and demand continue to materialize, related sectors may have greater upside potential, while overall progress on the application side could also drive the technology layer.
Editor/Rice
