The same confusion in China and the US: Can silicon-carbon evolve from differentiation to win-win?

Core takeaway: ① Stocks: extreme divergence between silicon-based and carbon-based. The divergence in U.S. stocks started earlier, and in China’s A-shares it is even more pronounced; the root cause is that silicon-based profits are clearly stronger than carbon-based. ② Real economy: the AI dividend has not yet benefited carbon-based industries. AI capital expenditure has become a highlight for the economies of both China and the U.S.; in China, token call volume surged by 81x, yet the consumer confidence index fell to the historical 5% percentile. ③ Outlook: silicon and carbon will ultimately reach a win-win outcome. When AI moves from large models to physical applications, China’s manufacturing will enjoy an engineers’ dividend and a computing-cost advantage, and then the integration of the old and the new economy will achieve win-win outcomes.

The most notable feature of the global economy and stock markets in the first half of this year is the divergence between silicon-based and carbon-based industries. This article focuses on the U.S. and China—the leaders of this new AI technology revolution—to analyze the underlying reasons behind the silicon-carbon divergence and to explore paths toward a future win-win outcome.

1. Stock market performance: two worlds of ice and fire

Judging by the share-price performance of silicon-based and carbon-based sectors, the divergence in U.S. stocks began earlier and is even more extreme in A-shares. Using CSI 300 and S&P 500 constituent stocks as samples, we divide them into two categories—silicon-based and carbon-based—and the divergent trajectory is clear: U.S. silicon-based sectors took the lead first, increasingly outpacing carbon-based starting in early 2024, with divergence intensifying after 2025. In contrast, in A-shares, silicon-based stocks only surpassed carbon-based in the second half of 2025, but then quickly widened the gap, especially in 26Q2’s two worlds of ice and fire.

Note: In A-shares, silicon-based companies include listed companies in CITIC primary-level industries such as communications, electronics, computers, power equipment and new energy; carbon-based companies include listed companies in 15 industries such as steel, building materials, construction, and transportation. Industries like large finance and machinery are classified as “intermediate.” In the U.S., silicon-based companies include listed companies in SIC secondary-level industries such as industrial and commercial machinery and equipment, measuring/analysis/control equipment, electronic equipment and components, communications services, and business services; carbon-based companies include listed companies in 40 industries such as food and related products, miscellaneous retail merchants, and oil and natural gas drilling. Industries like large finance and miscellaneous manufacturing are classified as “intermediate.”

Industry price moves also show that the divergence in A-shares is greater than in U.S. stocks. In the first half of 2026, in A-shares, about half of the 30 industries finished up; the top gainers were the two major silicon-based industries of electronics and communications. The building materials sector saw the largest gain mainly driven by AI-demand-induced orders for electronic boards. Among the 15 declining industries, 8 carbon-based industries fell by more than 10%, and at the single-stock level, the proportion of falling stocks is as high as 71%. By comparison, among the 54 SIC secondary-level industries of S&P 500 constituent stocks, 35 industries rose in the first half. While silicon-based industries did not post as standout gains as in A-shares, their overall driving power to the broader market was stronger, pushing more than half of U.S. stocks across the whole market to close higher.

By trading concentration, both A-shares and U.S. stocks have highly concentrated toward silicon-based. Taking CSI 300 and S&P 500 constituent stocks to divide into silicon-based and carbon-based categories, as of June 30, in A-shares, silicon-based sectors accounted for 37.7% of companies’ count and contributed 63.1% of trading value; in the U.S., silicon-based sectors accounted for 33.3% of companies’ count and achieved 67.9% of trading value. Among them, the heat of A-shares’ silicon-based sector increased more rapidly. In the U.S., silicon-based trading share has remained above 50% for a long time since 2021; it gradually rose after 2023. In A-shares, before 2023 its share was still below 35%, then it rose noticeably starting in 2025, and in the first half of this year it quickly broke through 60%.

The divergence in silicon- and carbon-based stock prices stems from differences in profitability. From the financial reports of listed companies, profitability in A-shares diverges clearly: in the silicon-based industries, year-on-year and quarter-on-quarter parent-company net profit growth is positive in the first quarter, with half of them growing by more than 30%, while most carbon-based industries lack growth momentum. In the U.S., profitability distribution is more balanced: while silicon-based industries lead, many carbon-based industries also show strong performance. From industrial enterprise data, in China, from January to May, industrial enterprise profits in the computer, communications, and other electronic equipment manufacturing industries grew year-on-year by 103.9%, far above the industry-wide 18.8%, but profits in 20 industries mainly carbon-based shrank by 15.4%. Meanwhile in the U.S., while silicon-based profits led in the first quarter, carbon-based sectors performed relatively steadily.

2. Real economy: uneven “hot and cold”

In the real economy, silicon-based and carbon-based sectors show uneven heat. For the China-U.S. economies driven by compute investment, the internal structure is diverging sharply: incremental demand comes mainly from silicon-based industries, while carbon-based demand is suppressed and even replaced.

Token call volume surged, but consumer confidence is weak. As AI applications deepen, 2026 is seeing a boom in real demand; currently China’s large-model token calls per week have reached 19.8 trillion, up 81x versus the same period in 2025. In the U.S., weekly calls reached 5.8 trillion, up about 3.9x year-on-year. By comparison, both China and the U.S. consumer confidence fell to historical lows: in May, China’s consumer confidence index was 89.9, at the historical 5% percentile; in the U.S. it was 44.8, the lowest since the index was launched.

The highlights of GDP growth in both China and the U.S. come from AI capital expenditure. In the U.S., the past boost to growth relied on consumption, but in this year’s first quarter, AI investment’s contribution to GDP growth reached 85%, exceeding consumption. China’s economic structure has strong external demand but weak internal demand, and the high export growth is mainly driven by U.S. AI capital expenditure. For electronic products, exports on average were boosted by about 11 percentage points year-on-year; this accounts for nearly 70% of the overall export increment, becoming a core source of China’s high export growth as AI capital expenditure spills over.

The core of the silicon-carbon split is that AI’s dividend is still circulating within silicon-based industries and has not spilled over to carbon-based fields yet. What AI reshapes first is corporate production functions and capital returns; improvements in residents’ income and end-demand have been relatively lagging. This leads to the current situation where silicon-based industries actively expand while carbon-based industries face passive pressure.

On the one hand, AI “arms race” itself requires sustained high-intensity spending, and it is still hard to spill over to the carbon-based economy. In how technology firms allocate profits, they are more inclined to put funds into compute, models, and data centers rather than immediately rewarding employees and shareholders. Taking the four largest AI companies in the U.S. as examples, capital expenditure is expected to rise to around 604 billion USD in 2026, nearly a doubling of 2024 and close to two times over, showing that in the AI era capital expenditure is clearly accelerating.

On the other hand, AI “substitution effects” are continuously squeezing employment and income expectations, creating a “second hit” to carbon-based industries. The shock of the new economy to traditional sectors is not only reflected in corporate profits and capital expenditure; it will also spread through employment. In the U.S. as an example, employment in high-AI-penetration industries such as technology and media, and manufacturing has remained sluggish and is significantly negatively correlated with the AI penetration index. With employment and income expectations weakening, residents’ consumption confidence naturally cannot rise.

3. Outlook: how to move from divergence to win-win?

A K-shaped divergence will not be the end state of the AI revolution. Looking back at past technological revolutions, the old and the new economies are not a zero-sum game; instead, they transform from early conflict and substitution into final integration and win-win outcomes. Ultimately, carbon-based industries represent people’s fundamental needs, and the productivity boom brought by the silicon-based revolution must still return to people’s needs. In the future, as AI costs fall and application scenarios become more abundant, the silicon-based dividend will bring widespread demand expansion, driving silicon and carbon from “divergence” toward “win-win.”

In theory, new technology always first transforms supply and then creates demand. Early efficiency gains shift the supply curve to the right; but as costs decline and scenarios expand, new demand will gradually take over, shifting the demand curve to the right and forming a new dynamic equilibrium with both quantity and price rising together. For example, in the early days of the internet revolution, the U.S., as the origin of the technology, benefited first; later, infrastructure investment exploded, and the dividend spilled over along the industrial chain and trade chain. The internet revolution accelerated globalization; after China joined the WTO in December 2001, it absorbed the dividend, and China’s share of global GDP scale changed upward. In 2003, China achieved a switching of the new and old drivers of growth, and corporate ROE continued to improve—China’s assets were comprehensively revalued.

When AI penetrates from the digital world into the physical world, silicon-based and carbon-based will achieve win-win. AI is currently still in the stage of compute, models, and infrastructure first. The U.S. is both the origin and the frontrunner. However, once AI shifts from large models to physical applications—accelerating adoption of AI cars, AI phones, and robots—China’s ability to absorb the silicon-based revolution will become explicitly visible.

First, China has a huge engineers’ dividend. In 2025, the average annual salary of IT industry engineers in China is about 198k USD, much lower than the U.S.’s roughly 180k USD. This means that under the same conditions, China can support larger-scale engineering teams, higher-frequency product iterations, and faster manufacturing rollout.

Second, China has a large computing-cost advantage. The token price of China’s mainstream large models is only 1/5, or even lower, than that of comparable models in the U.S. This means that when AI is deployed at scale, the call threshold is lower, making it easier to spread and become more widely adopted. When competition shifts from pure compute and algorithms to low-cost, high-frequency, large-scale applications, China’s advantage will be further amplified.

Intelligent manufacturing represented by intelligent vehicles and robots is expected to become the bridge connecting silicon-based and carbon-based sectors, driving growth in residents’ income and consumption; at that time, incremental demand will diffuse toward carbon-based industries. Then, China’s economic transition will move from quantitative change to qualitative change, and the scale of the new economy will exceed that of the old economy. For comparative analysis between the old and new economies, earlier reports have covered this—see 《新老经济的力量对比：股市和实体维度》、《PPI转正上下半场逻辑和股市表现不同：借鉴98-03年》.

Source: Sun Yu-gen’s thinking

Risk warning and disclaimer terms

        There are risks in the market; invest prudently. This article does not constitute personal investment advice, and it does not take into account any specific users’ special investment objectives, financial situations, or needs. Users should consider whether any opinions, viewpoints, or conclusions in this article align with their specific circumstances. Invest at your own risk and bear responsibility for the results.