AI News

Li Xunlei: With external demand declining and multiple constraints, how to stabilize growth in 2027

HuxiuUpdated Nov 13 21:59

AI summary

Li Xunlei and Li Chenxi judge that the probability of China's export growth rate declining in 2027 has risen substantially: the growth rate of US AI capital expenditure will decline, compounded by the high base in 2026 and falling prices of AI products. In 2025, net exports contributed 32% to GDP growth; weakening external demand would deal a major blow to stabilizing growth, while fixed asset investment has already fallen for two consecutive years and retail sales grew only 1.1%. Policy may rely more on central government leverage, with the general budget deficit ratio at about 4%, ultra-long special treasury bonds possibly at 1.8–2 trillion yuan, new special bonds possibly reaching 5 trillion yuan, and a comprehensive rate cut unlikely.

Body · AI translation

[[**In 2027, export growth is highly likely to fall back: the slowdown in US AI capital expenditure and the high base will weaken external demand; expanding domestic demand becomes key to stabilizing growth, but local fiscal constraints, household deleveraging, and income and social security constraints are prominent, and policy may rely more on central government leverage and structural tools.**

**Key points**

**1. Support from external demand will weaken markedly**

The growth rate of US AI capital expenditure may decline, and the high base in 2026 and falling prices of AI products raise the probability of an export pullback; in 2025, net exports contributed 32% of GDP growth.

**2. Expanding domestic demand is constrained by fiscal and household factors**

Fixed asset investment has fallen for two consecutive years, with local governments constrained by declining land-sale revenue and debt resolution and arrears clearance; retail sales grew only 1.1%, and household deleveraging and income and social security difficulties suppress consumption.

**3. Fiscal leverage to increase, monetary policy to remain structural**

The general budget deficit ratio in 2027 may be about 4%, ultra-long special treasury bonds may be 1.8–2 trillion yuan, and new special bonds may reach 5 trillion yuan; a rate cut is unlikely, and PSL may be expanded to stabilize infrastructure.

**4. Consumption, livelihoods and AI investment need to be weighed**

This article is from the WeChat public account: Li Xunlei Finance and Investment, authors: Li Xunlei, Li Chenxi

Looking back at 2026, the biggest feature of the global economy was that it unfolded around the substantial growth of US AI capital expenditure, with prices of AI-related products rising sharply—this is precisely where China's opportunity lies. In the first three quarters, China's exports grew 21.3% year on year, of which integrated circuits raised export growth by about 5 percentage points, and the growth came almost entirely from price increases. However, China currently benefits mainly from hardware manufacturing, not from large models and applications.

However, the probability that export growth will fall back in 2027 has risen substantially. The growth rate of US AI capital expenditure will decrease, and the export base in 2026 is also very high. In 2025, 32% of GDP growth came from net exports; once external demand weakens, the impact on stabilizing growth will be large. Expanding domestic demand is likewise subject to many constraints: fixed asset investment has fallen for the second consecutive year, and local governments are constrained by declining land-sale revenue and the pressure of debt resolution and arrears clearance; retail sales growth was only 1.1%, the household sector is deleveraging, and the two major difficulties of income and social security remain to be resolved.

Looking ahead to 2027, the policy intensity for expanding domestic demand should in theory be greater than in 2026. On the fiscal side, it will mainly rely on increased leverage by the central government; ultra-long special treasury bonds may be expanded to 1.8–2 trillion yuan, and new special bonds may reach 5 trillion yuan. Monetary policy will still focus on structural tools, and a comprehensive rate cut is unlikely; the most likely move is to continue expanding the scale of PSL so that infrastructure investment stops falling and stabilizes. But one cannot have one's cake and eat it too: to balance fiscal input in promoting consumption, safeguarding livelihoods, and AI and other areas, the central government will need to further increase borrowing.

US capital expenditure growth may fall sharply next year

Looking back at the soon-to-pass 2026, the biggest feature of the global economy was that it unfolded around the substantial growth of capital expenditure in the US AI sector. How large the scale of capital expenditure actually is varies across statistics; for example, some institutions estimate that US hyperscale cloud providers' capital expenditure in 2026 will reach $870.5 billion, and forecast that it will soar to $1.36 trillion in 2027, accounting for 4.1% of GDP.

Because the substantial increase in AI capital expenditure has driven up demand upstream in AI, the growth rate of global AI-related product and equipment manufacturing and the growth rate of import and export trade have also risen. For example, Taiwan, China, where the electronics industry is highly developed, saw real GDP grow 14.22% year on year in the first three quarters of 2026, while nominal GDP growth was as high as 20.46%.

However, while AI drives US economic growth, it also brings negative effects. First, inflation is also somewhat correlated with AI. For example, AI development requires building large numbers of data centers, which are concentrated in the PJM grid region; power construction cannot keep up with electricity demand growth, and capacity auction prices have risen to 11.5 times the 2024/25 level, yet still fail to purchase enough capacity. Capacity costs will ultimately be spread across all users' electricity bills, so the electricity cost of AI spreads to ordinary businesses and residents, becoming a channel through which AI affects US inflation.

PJM capacity auction clearing prices: data center electricity demand drives power prices soaring (USD/MW-day)

Another negative effect brought by AI is employment. The upstream of the US AI industry inherently creates very few jobs, and since 2025, after the improvement of US AI programming capabilities, employment among software developers aged 22 to 25 has fallen by about 20% compared with November 2022, while that of those aged 35 to 40 has instead risen by about 11%. This shows that AI first replaces entry-level positions, and young people are hit harder. In the future, as AI applications continue to expand, their characteristic of substantially replacing human labor will become more pronounced.

In addition, the development of the AI industry has led to intensified income distribution divergence in the United States. Since 2022, the ratio of US household financial assets to disposable income has risen from 5.5 times to 6.5 times, with the increase coming almost entirely from stocks. The wealth effect of the AI rally is mainly concentrated in high-income households that hold more stocks, further intensifying K-shaped divergence. Over a longer period, wealth concentrating in the hands of a few is a recurring phenomenon in market economies, and AI has accelerated this process. If unemployment rises markedly because of AI, consumption will weaken accordingly, social pressure will also rise, and ultimately this may prompt companies to cut capital expenditure.

Therefore, the Democrats' landslide victory in the US midterm elections was not essentially because the Democratic Party is deeply popular, but because the public expects the Democratic Party to curb the excessively rapid development of AI in the United States, since these developments have already caused consequences such as inflation, fewer job opportunities, and a widening wealth gap.

Even without the midterm election factor, the growth rate of US AI capital expenditure will decline in 2027. The midterm elections have brought more uncertainty and investment risk to US AI companies; therefore, a decline in the growth rate of capital expenditure will not only slow US economic growth but also lead to a decline in the growth rate of global AI-related trade volume.

Behind China's strong export growth this year:

In the first three quarters, China's cumulative exports grew 21.3% year on year (in USD terms), and cumulative imports grew 27.5% year on year. The reason China's imports and exports could grow substantially is closely related to the sharp increase in China's exports of electronic products; for example, integrated circuits raised the national export growth rate by about 5 percentage points, and excluding them, other exports grew by about 16%. Although the quantity of integrated circuits grew by only 3.2%, the average price rose by 101.4%, meaning that almost all of the growth came from price increases, with memory being the category with the largest increase.

For example, from the 14th Five-Year Plan to the 15th Five-Year Plan, AI's positioning in policy has risen from an emerging industry to infrastructure supporting economic operations, and government investment has continued to increase. Government guidance funds have already invested about USD 184 billion in AI, and in 2025 a new national venture capital fund of nearly RMB 1 trillion was established, far exceeding private capital in total; however, leading large models are mainly cultivated by private enterprises and industrial capital, with state-owned capital mostly following up in later stages.

From the corporate level, the capital expenditure of large internet companies has clearly accelerated in the past two years, but the overall scale is still far smaller than that of US cloud vendors, and the pace of expansion is also constrained by chip supply. Overall, the scale of government funding is large, while frontier innovation mainly relies on private enterprises; private capital follow-up is limited, mainly because AI's profit model is not yet clear, and application scenarios and willingness to pay are still insufficient.

China's average IC import price, IC output, and semiconductor equipment imports year on year

Source: WIND, Zhongtai International

In terms of beneficiaries, China currently mainly benefits from hardware manufacturing rather than large models and applications. In the third quarter of 2026, exports of integrated circuits and automatic data processing equipment grew 89% year on year, far higher than the 21% growth rate of total exports; the profit margin of the electronics manufacturing industry rose to 6.8%, the highest since 2019; while the profit margin of the software industry remained at around 12%, with no obvious improvement.

China's exports year on year: total vs integrated circuits + automatic data processing equipment (%, monthly average within the quarter)

Note: Bar values are the simple average of monthly year-on-year growth rates within the quarter; January-February data for 2025Q1 are partially missing, and integrated circuits + automatic data processing equipment only include March.

Source: WIND, Zhongtai International

Profit margins of China's electronics manufacturing and software industries (%, monthly average within the quarter)

Although there are many uncertainties in 2027, the probability of a decline in export growth has risen substantially. Looking back at cumulative export growth from 2020 to 2025, China's cumulative exports grew 45.6%, while the global figure was 48.8%, meaning China was 3.2 percentage points below the global level. The main reason for being below the global level is the sharp decline in the export price index, even though export volume growth was not low. But 2026 should be an exception, as chip export prices rose sharply, causing total export growth to outperform global export growth.

The reason for predicting a decline in export growth in 2027 is, first, that the export prices of AI-related products will fall significantly, thereby leading to a decline in export value growth, because 2026 is the first year of the full-scale explosion of the AI industry, and the export base that year is very high, causing growth to fall back in 2027; the second reason is that the growth rate of US AI capital expenditure will decline, leading to a decline in the growth rate of global AI product trade volume, and in turn a decline in the growth rate of export demand for Chinese goods.

In fact, during 2025-2026, when Trump sharply raised tariffs on China, China's export growth actually rose substantially and made a large contribution to GDP growth. For example, GDP growth in 2025 was 5%, of which 32% came from net exports, the largest contribution in a year since 1998. In the first three quarters of this year, the contribution of net exports to GDP growth was about 26%, a decline in contribution rate, mainly because imports grew faster.

Data show that from 2010 to 2019, the average contribution of China's net exports to GDP growth was negative (-1.3%). But from 2024 to the present, the average contribution rate of China's net exports to GDP growth has been close to 30%. After entering 2020, the contribution of exports to economic growth has clearly increased. On the one hand, this depends on the growth of external demand; on the other hand, it also reflects the growing advantages of Chinese manufacturing globally and stronger supply capacity.

The above data show that if external demand weakens in 2027, the impact on China's efforts to stabilize growth will be significant. Looking at China's economic data for the first three quarters, the growth rate of fixed asset investment was -7.2%, already the second consecutive year of decline (2025 saw the first negative reading since statistics began). On the one hand, this is dragged down by the downturn in the real estate cycle; on the other hand, it is because the growth rates of manufacturing and infrastructure investment have also begun to fall.

Year-on-year trends of private and state-owned fixed asset investment

Source: WIND, Zhongtai International

From January to September this year, the growth rate of manufacturing investment fell by 2.1%, while the growth rate of infrastructure investment fell by 6.1%; the growth rate of private investment was -10.4%, and the growth rate of investment by state-controlled entities was -3.2%. If the pace of private investment growth depends on how high private firms perceive the expected return on investment to be and how large the risks are, then against the backdrop of strong supply and weak demand, it is still understandable that the willingness to invest in real estate and manufacturing has declined—but why have both infrastructure investment and state-owned investment shown negative growth?

Perhaps an answer can be found from the perspective of the timing progress of fiscal expenditure. In the first three quarters, the growth rate of general budget expenditure was only 1%, clearly slower than the 3% growth rate in the same period last year, while the full-year target growth rate is 4.4%; at the same time, using the revenue-expenditure gap in the first and second fiscal accounts as the broad deficit measure, this year's utilization progress is about 63.6%, lower than 74.5% in the same period last year. Why is fiscal budget expenditure below the expected target?

The main reasons lie in two aspects. First, local governments' revenue from land sales has fallen sharply, but the related expenses are rigid. For example, from January to September, the national government fund budget revenue was only 2.50 trillion yuan, down 18.6% year on year; government fund budget expenditure, however, reached as high as 6.28 trillion yuan, with the revenue-expenditure gap reaching 3.78 trillion yuan. Land transfer fee revenue has fallen sharply, while rigid expenditure has not declined much, reducing local governments' supporting funds for infrastructure investment.

Second, local governments face increasing pressure from debt resolution and clearing arrears, because June 30, 2027 is the deadline for local governments to stop relying on platform company financing and to complete the clearing of arrears, so relevant arrangements need to be made starting this year.

In the first three quarters of this year, the utilization progress of the broad deficit was lower than in the same period in recent years

Source: WIND, Zhongtai International

From the growth rates of general budget expenditure at the local and central government levels, it is not hard to see that the growth rate of local governments' own-level expenditure was only 0.2%, while the growth rate of central government expenditure was 5.8%. Therefore, to rely on investment to drive the achievement of the stable growth target—in the past this depended on local governments actively responding; now, constrained by local fiscal capacity, they have the will but not the strength.

Looking at the horse of domestic demand—consumption—in the first three quarters, the growth rate of retail sales of consumer goods was only 1.1%. Although in China's 2026 consumption promotion policies the quota allocated by the central government for trade-ins is 250 billion yuan, the consumption growth rate of trade-in categories of goods is precisely the category that has fallen the most. For example, the month-on-month year-on-year retail sales of five categories above designated size—home appliances, furniture, cultural and office supplies, automobiles, and communication equipment—fell from -9.3% in August to -10.5%. The forces that used to drive consumption have now become a drag.

The decline in retail sales of trade-in categories continues to widen

Source: WIND, Zhongtai International

Service consumption, by contrast, is relatively stronger. In the first three quarters, retail sales of services grew 5.0% year on year, 4 percentage points higher than retail sales of goods, among which travel consulting and leasing services and cultural, sports and leisure services grew by 10.7% and 9.7%, respectively. However, after all, the volume of service consumption is limited, and it is still difficult to offset the weakness of goods consumption.

Consumption constraints are mainly reflected in two aspects: supply constraints and demand constraints. From the supply side, basic needs such as food, clothing, housing and transportation have all been met. For example, according to 2023 data from the Food and Agriculture Organization of the United Nations (FAO), the per capita daily calorie supply in mainland China was about 3,527 kilocalories, ranking 23rd among 200 countries and regions, far exceeding Japan, South Korea and China's Hong Kong, Macao and Taiwan regions.

From the property market perspective, in the first three quarters of this year, nationwide sales of newly built commercial housing were 5.37 trillion yuan, down 13.9% year on year; in 2027 the property market will most likely continue its downward trend. Although since 2021 the total supply of newly built commercial housing has been decreasing, the supply scale in the second-hand housing market is extremely large, and weak demand means the pattern of oversupply has not fundamentally changed, which also leads to insufficient demand for many real estate-related upstream and downstream consumer goods.

In transportation, the total mileage of high-speed rail, expressways and subways all leads the world by a wide margin. In the first three quarters of this year, the growth rate of domestic automobile sales fell by 22.4%. At present, new subway projects nationwide have not been approved for two years, and the situation of strong supply and weak demand is also difficult to change.

From the demand side, the household sector began deleveraging in 2024, and this year the balance sheet contraction is obvious. In the first three quarters, the balance of household loans decreased by 1.03 trillion yuan, the first such occurrence since 2000, indicating that the consumer confidence index is still weakening. Although the household sector's debt service ratio (quarterly principal and interest payments/disposable income) has declined somewhat, in international comparison it still substantially exceeds that of the United States and Japan.

Another constraint on demand is the two major problems of household sector income and social security. Although statistical data show that the growth rate of residents' disposable income is in step with GDP growth, there may be structural problems, namely that the income growth rate of low- and middle-income people is lower than the average growth rate. For example, in the first three quarters of 2026, the growth rate of the national per capita disposable income average was 5.3%, while the median growth rate was 4.6%, with the median only 82.9% of the average.

The main consumer group is the low- and middle-income class. The vast majority of China's flexible workers belong to the low- and middle-income class, and there is also a large agricultural population; they generally face the problems of insufficient social security coverage or low social security pensions.

In summary, against the backdrop of expected decline in external demand, expanding domestic demand has undoubtedly become an inevitable choice. But what needs attention is that over the past five years, expanding domestic demand has consistently been the focus of China's macro policy, yet investment growth and consumption growth have still shown downward trends. In other words, expanding domestic demand has already been constrained by various factors, and no growth model can work all the time.

In 2027, expanding domestic demand needs more force

At the State Council executive meeting on November 6, the following matters were proposed: First, the pace of construction and implementation of major projects and the "six networks" should be accelerated, which indicates that the progress of manufacturing and infrastructure investment in the first three quarters was indeed lower than in previous years, and it is estimated that the pace can be accelerated in the fourth quarter to achieve the full-year GDP growth target; second, it is necessary to "facilitate a virtuous cycle between investment and consumption"; third, solid work should be done in stabilizing employment, preventing risks, and ensuring people's livelihood; fourth, risks in areas such as local government debt and small and medium-sized financial institutions should be continuously resolved.

In another month, the Central Economic Work Conference will be held to decide the direction and intensity of fiscal, monetary and other economic policies for 2027. It is estimated that in 2027, macro policies will still revolve around the above matters. Since external demand may decline, then in theory the policy intensity for expanding domestic demand should be greater than in 2026.

However, fiscal, monetary policies and policies in subdivided fields will likewise be constrained by various parties. As far as fiscal policy is concerned, the broad fiscal budget expenditure for 2026 is already close to 54 trillion yuan, accounting for about 37% of 2026 GDP (assuming China's GDP scale in 2026 is 145 trillion yuan, the same below), while the broad fiscal deficit ratio is about 8%, so the proactive fiscal policy is indeed very proactive.

To achieve the goal of stabilizing growth, it mainly relies on the central government to add leverage, because the central government's leverage ratio is currently more than 30%, and there is still room. It is estimated that in 2027 the general budget deficit ratio will remain at around 4%, and the scale of ultra-long-term special government bonds may expand to 1.8-2 trillion yuan.

In addition, the scale of special bonds used to ease the cash flow pressure of local governments still needs to be expanded, especially after urban investment platforms can no longer provide various forms of indirect funding support to local governments. The scale of newly added special bonds in 2026 is 4.4 trillion yuan, and it is estimated that it may reach 5 trillion yuan in 2027.

Monetary policy will still be dominated by structural monetary policy, and the possibility of comprehensive interest rate cuts is not large. Although bank loan growth has declined in recent years, reflecting a reduction in market-based indirect financing demand, a systemic rate cut would lead to a narrowing of banks' net interest margins and bring greater pressure on bank operations. Because in this round of the real estate downturn cycle, the pressure borne by banks is the greatest. This is why the central government has repeatedly emphasized the risks of small and medium-sized financial institutions.

In 2025-26, the central government has already injected 800 billion yuan into large state-owned banks and insurance institutions, and local governments have also injected capital into local banks or become major shareholders (the amount is unknown). The purpose is to enhance the capital adequacy and risk resistance of banks and other financial institutions.

Although structural monetary policy tools can also reduce the cost of financing entities, since May 2025 the policy rate has stayed at 1.40%, and there has been no further reserve requirement ratio cut; structural tools carry out "targeted price cuts" by field, without compressing the interest margin on all loans. Before June 2025, PSL was the main tool of structural monetary policy, and by the end of August its balance had dropped to 0.63 trillion yuan; it has been replaced by central bank relending, which is mainly directed toward the "five major articles of finance," with a balance of 4.6 trillion yuan by the end of June.

PSL balance and monthly changes (100 million yuan)

Source: WIND, People's Bank of China, Zhongtai International

It is worth noting that at the end of September the PSL tool was reactivated, with the interest rate lowered to 1.5%, and the "six networks" were included in the scope of support, using policy banks to invest directly in infrastructure to make up for the insufficient leveraging power of relending. But overall, the scale of funds provided by structural policy tools is currently estimated at 5-6 trillion yuan, which is only about 2% compared with the current loan balance of about 280 trillion yuan, making it difficult to drive down the financing costs of the whole society.

Against the backdrop of deleveraging by the household sector and private enterprises, credit expansion is already being supported by the government. The proportion of government bonds in total social financing rose from about 20% at the beginning of 2020 to 38%. The main channel of money creation has shifted from bank credit to government bond issuance, indicating that what is truly expanding demand is fiscal policy, while monetary policy only cooperates, for example, "relending + interest subsidies" is a combination of the two.

Government bond financing and its proportion in newly added total social financing

Although some core cities' real estate markets show signs of stabilization, the overall downward trend remains unchanged. For the housing market to stop falling and stabilize, a comprehensive interest rate cut should be a relatively effective means, but the risk is that it is unknown how effective it will actually be. For example, in 1996 the Bank of Japan cut the interest rate to 0.5%, and in 1999 cut it to zero, neither of which achieved the goal of stabilizing housing prices.

Therefore, the most likely approach is to continue expanding the scale of PSL and, by accelerating the process of laying out the "six networks," bring infrastructure investment growth to a halt and stabilize it. In other words, the moderately loose monetary policy in 2027 will still be reflected as structural monetary policy, and the scale of low-interest financing it provides will be further expanded.

In 2027, in terms of investment in the "two major" areas, it is estimated that the main line will be the comprehensive implementation of the "AI+" initiative. Promote the transformation of traditional industries by artificial intelligence, and accelerate the application of new-generation intelligent terminal scenarios such as intelligent connected new energy vehicles, AI mobile phones and computers, and humanoid robots. In this way, manufacturing investment will be driven from negative to positive.

But what needs to be considered is that in the United States, AI investment is constrained by electricity, manifesting as rising electricity prices; China's electricity and manufacturing capacity are relatively sufficient, and the main constraint is on the demand side. AI applications further enhance supply capacity, and in China they are more likely to manifest as price competition. The prices of large-model services continue to decline, and related manufacturing capacity continues to expand, while corporate profits and labor income may not grow in sync. If AI further replaces some jobs and lowers the share of labor income, consumption will also be suppressed. Overall, AI applications may increase rather than ease the downward pressure on prices, which is clearly different from the United States.

What is worth looking forward to in 2027 is "formulating and implementing a plan to increase urban and rural residents' incomes," which is a goal proposed at the Central Economic Work Conference at the end of 2025. If it can be delivered in 2027, it will help boost consumption.

But considering that there are many constraints on fiscal budget expenditure, and one cannot have both fish and bear's paw, it is unrealistic to substantially increase spending on both boosting consumption and promoting investment. Moreover, AI is a long-term investment, while livelihood spending is continuous investment with rigid annual growth, and excessive AI investment may crowd out livelihood spending. To take into account both expanding consumption and long-term investment in AI, it is necessary for the central government to further increase borrowing.

In summary, the external environment facing 2027 will be more severe than in 2026, and internal structural and cyclical pressures still show no signs of weakening, so greater efforts are needed to expand domestic demand. Of course, there are always more solutions than difficulties, and I believe that in 2027 China will have the ability to overcome various challenges and strive to achieve the goal of stabilizing growth. Channel: Finance and Economics. This content comes from the Internet and the original link. The views only represent the author himself and do not represent the position of Huxiu. If there is any copyright issue, please contact hezuo@huxiu.com, and we will verify and handle it promptly. Everyone who is changing and wants to change the world is on the Huxiu APP[ ]]]

Source

Huxiu