Core conclusions: 1) AI is currently in the "creative destruction" phase on the left side of the J-curve, with an obvious crowding-out effect. 2) On the employment side, China shows silicon-based labor crowding out carbon-based labor, while the US shows capital crowding out labor. 3) On the funding side, whether through equity, bonds, or credit, both China and the US are accelerating their focus on silicon-based sectors. 4) On the policy side, AI has become a core arena of great-power competition. 5) When AI moves into the physical world, it will reach the "prosperity inflection point" on the right side of the J-curve, when China's advantages will become prominent and asset prices will be revalued.

The AI crowding-out effect: taking China and the US as examples. Chief Economist of Guosen Securities Xun Yugen, practicing qualification number: S0980525090001. Chief Macro Analyst of Guosen Securities Tian Di, practicing qualification number: S0980524090003. Report release date: September 29, 2026. The surging AI revolution has sparked humanity's imagination about the future and also driven a new global wave of capital. However, what most ordinary people feel first at present is greater pressure to survive. Why has this situation emerged? Because AI is still in the "creative destruction" phase in the early part of the "J-curve." This article takes China and the US as examples to analyze the current crowding-out effect of AI on other sectors.

1. Employment: Both volume and price are rising in the AI sector. The "quantity" of jobs is the most intuitive perspective for observing AI's crowding-out effect. Technological revolution is essentially a repricing of people. When the relative demand for silicon-based and carbon-based labor changes, the job market will seek a new equilibrium through changes in positions and wages, and ultimately labor will migrate from old growth drivers to new ones. China's employment-side crowding-out effect is already quite obvious, with silicon-based expansion and carbon-based contraction. Over the past four years, AI-related "silicon-based" employment has grown at a compound annual rate of 3.2%, while typical "carbon-based" industries such as construction and utilities have fallen by 11.0% and 1.5%, respectively. More importantly, the silicon-based sector is still small, and its employment expansion is hard to offset the larger-scale decline in carbon-based employment. From 2023 to 2025, China's AI-related employment increased by about 2.7 million, while non-AI-related employment decreased by about 9.8 million. In the US, the employment crowding-out effect of silicon-based sectors on carbon-based sectors is reflected at the firm level. US carbon-based employment is still growing, but silicon-based employment has instead contracted: over the past four years, US information technology employment has fallen at a compound annual rate of about 3.0%, while construction and utilities grew by 1.7% and 2.1%, respectively. From the production function F(K, L), China is mainly undergoing a restructuring within the labor factor, that is, silicon-based L crowding out carbon-based L, while in the US it occurs between factors, that is, capital K crowding out human labor L. This can be corroborated by the number of business registrations. From 2022 to now, the number of employees in the US information technology industry has fallen by 11.9%, while new business registrations rose from 7,920 to 13,200, an increase of 66.4%, hitting a record high. The "price" of labor returns is another important dimension for observing crowding out. Labor returns are the "cause" of the crowding-out effect. When demand for AI-related positions continues to expand, talent scarcity pushes up wage returns, and the higher wage premium further attracts labor inflows, forming a positive cycle. Both China and the US show larger wage increases in AI fields. In 2024, the total wage growth of China's AI-related industries rose from 5.2% to 8.5% year on year, while other industries rose only from 3.3% to 3.5%, and the growth gap between the two widened from 1.9 percentage points to 4.9 percentage points. The US is similar. As of July 2026, weekly wages in AI-related industries were up 4.8% year on year, higher than the overall 3.5%. The price signal of wages has already affected educational choices. The admissions advantage of AI majors at China's 985 universities continues to expand, with the relative leading rank in the college entrance examination widening from 782nd in 2022 to 2,600th in 2025, representing that candidates are willing to pay a higher "score premium" for AI majors. The same is true in the US. Over the 10 years from 2014 to 2024, the number of computer and information science degrees awarded doubled from 56,000 to 122,000, while social sciences fell to 162,000 over the same period.

2. Capital: Funds are accelerating toward AI. Capital is another key factor supporting the expansion of the AI industry. In the short term, when total social financing changes little, rising capital demand in silicon-based industries will inevitably crowd out traditional industries to some extent. At present, AI competition between China and the US continues to intensify. The US holds a leading advantage with large models, computing power, and platform ecosystems, while China is still in a catch-up stage and needs even more to "concentrate resources to accomplish major tasks," filling shortcomings through extraordinary investment in capital and credit. From the perspective of equity financing, the share of silicon-based IPO fundraising has risen sharply. The proportion of IPO fundraising by silicon-based industries in the A-share market has increased significantly since 2021, reaching a historic high of 53.5% so far this year, while the US stock market is as high as 79.3%, showing an obvious trend of focusing on silicon-based sectors. At the same time, the share of silicon-based IPOs in both countries has not risen substantially, indicating that silicon-based companies are raising larger amounts per project, and capital is accelerating toward the leaders. From the perspective of debt financing, silicon-based sectors' ability to attract funds has strengthened. In direct financing, the allocation weight of bond funds to the AI industry chain is increasing. The share of silicon-based corporate bond issuance in total non-financial issuance in both China and the US has gradually risen from lows in 2024. Compared with the US, China's silicon-based companies rely relatively less on bond financing, mainly because many related companies are still in an early stage of development, with constraints on business models and earnings stability. With the development of the silicon-based industry itself and the acceleration of building the bond "technology board," China, as the world's second-largest bond market, is expected to provide stronger support for financing technology companies. In indirect financing, bank credit resources are further tilting toward the technology sector. Loan balances for China's high-tech enterprises and technology-based small and medium-sized enterprises have maintained double-digit growth, significantly faster than infrastructure, services, and industry. Since 2025, large syndicated credit lines in the US technology and business services sector have also accelerated markedly, leading other key areas such as healthcare, manufacturing, and materials.

3. Policy: AI is the core point of great-power competition. AI has become the core battlefield of global gaming, and resource input has gone beyond traditional market logic. Facing the US's first-mover advantage and technological containment, China must proactively concentrate more fiscal, legislative, and industrial resources on AI-related fields, accelerating the catch-up in the spirit of the "Two Bombs, One Satellite" project. First, at the budget level, public resources are tilting toward scientific and technological research and development. Over the past four years, the share of basic research funding in China's R&D has continued to rise, reaching 7.1% in 2025. Although there is still an obvious gap with scientific and technological powers such as the US, investment in basic research is accelerating further on the basis of the expansion of total R&D investment, creating a certain resource crowding-out effect on applied research and even carbon-based fields. For example, compared with 2020, in 2024 the share of the Natural Science Foundation in science and technology expenditures in the general public budget increased by about 0.24 percentage points, while the share of the Social Science Foundation fell rather than rose. Second, the focus of policy legislation is tilting toward AI. From 2017 to 2025, China's central regulations involving "artificial intelligence" increased from 4 to 189, and local regulations increased from 46 to 599. At the same time, however, the total number of all central and local regulations fell from about 359,000 to 212,000, a decline of about 41%, and the proportion of AI-related regulations rose from about 0.14 per mille to about 3.7 per mille. It is relatively rare for the US Congress to legislate support for a specific industry, but Trump, as a Republican populist leader, placed AI competition at a relatively high priority after taking office, and since 2025 there have already been 9 executive orders directly increasing support for AI and its infrastructure. Third, at the subsidy level, both China and the US are using "real money" to compete for AI and advanced manufacturing. The scale of China's tax and fee subsidies supporting scientific and technological innovation and high-end manufacturing development rose further from 2.63 trillion yuan in 2024 to more than 3.51 trillion yuan in 2025. The US is also continuing to increase tax incentives, with the federal research and development tax credit rising from about $36.1 billion in 2025 to an estimated $38.8 billion in 2027. The US government is also further integrating federal research facilities, computing power, data, and private capital through bills such as the "Genesis Mission," "Stargate," and the "AI Action Plan," directing government resources and corporate investment together toward key areas such as AI and chips.

4. Summary: Supply creates demand, and silicon and carbon will eventually achieve win-win outcomes. In the "creative destruction" phase on the left side of the J-curve, the crowding-out effect is greater than the incremental effect. From employment and capital to policy, social resources at the current stage are more biased toward silicon-based fields, which is the core reason for the divergence between macro data and micro perceptions. But this does not mean that "silicon-based" is a "deflationary revolution." When AI moves from the virtual world to the physical world, the technological dividend will radiate from the "self-circulation" of a few silicon-based sectors to broader industries such as automobiles, manufacturing, and services. New supply will further create new demand, and silicon and carbon are also expected to move from "substitution" to "win-win." In the future, the "prosperity inflection point" on the right side of the J-curve will arrive, with silicon and carbon achieving win-win outcomes. In previous reports, we analyzed that the AI revolution will ultimately move into the physical world. When AI redefines manufacturing, supply creates demand, total demand will expand, and intelligent manufacturing will replace industrial manufacturing. By then, China's advantages will become prominent. See "The Same Confusion for China and the US: Can Silicon and Carbon Move from Divergence to Win-Win - 20260710" and "How to Save Domestic Demand? - 20260827." First, China has a huge engineer dividend. In 2025, the average annual salary of IT engineers in China was about $34,000, significantly lower than about $180,000 in the US. This means that under the same conditions, China can support a larger-scale engineering team, more frequent product iteration, and faster manufacturing implementation. Second, China has a huge computing power cost advantage. The token price of China's mainstream large models is only one-fifth or even lower than that of comparable US models. This means that when AI is deployed on a large scale, its usage threshold is lower, making diffusion and popularization easier. When competition shifts from pure computing power and algorithms to low-cost, high-frequency, large-scale applications, China's advantages will be further amplified. Intelligent manufacturing represented by smart cars and robots is expected to become the link connecting silicon-based and carbon-based sectors, driving growth in household income and consumption, and by then incremental demand will spread to carbon-based sectors. At that time, China's economic transformation will move from quantitative change to qualitative change, with the size of the new economy surpassing the old economy, that is, China's "2003 moment" recurring, corporate ROE rebounding, and Chinese asset prices undergoing comprehensive revaluation. Risk warning: AI development and application progress may fall short of expectations.