
A new Citi Research report from a team led by Chief China Economist Xiangrong Yu explores how AI has become the defining force shaping China’s K-shaped economy and looks at the divergence between the new economy — the strong leg of the K — and the weaker leg that represents traditional sectors. These fault lines, in our view, are being deepened by the AI supercycle.
We have described China’s post-pandemic recovery through two persistent themes: Supply has continued to outpace domestic demand, with exports filling much of the gap, and the new economy has consistently outperformed the old. Early 2025’s “DeepSeek moment” accelerated that trend by reinforcing confidence in China’s AI capabilities; since then, AI adoption has expanded rapidly across the economy, increasing its influence on growth, investment, and market performance.
The AI supercycle is powering production, exports, and new-economy activity, reinforcing the strong leg of the K, while concerns about future labor displacement continue to weigh on consumer confidence, despite the fact that measurable effects remain limited so far. AI-related industries, cities, and firms are increasingly pulling ahead, while many traditional sectors continue to struggle, and the AI buildout may divert investment away from those sectors.
We maintain our 2026 GDP growth forecast of 4.7% year over year. We expect AI-related activity to remain the main anchor for growth, particularly through supply-side strength. While the second quarter is likely to represent the low point of the year due to the Middle East shock and delayed fiscal deployment, we expect conditions to improve in the second half. On prices, the first phase of producer-price reflation was driven by energy, but we believe AI-related demand and anti-involution policies could provide the next source of support. As a result, nominal growth could reach a five-year high of 6.7% in 2026.
The AI economy is rapidly moving from software development toward large-scale infrastructure deployment, with several indicators illustrating the speed of the shift. China’s daily token usage surged to 140 trillion in March from just 100 billion at the start of 2024. Intelligent computing capacity tripled in 2025 vs. 2024. Equity-market leadership has also rotated from internet platforms toward semiconductor, hardware, and materials companies, reflecting growing investment in AI infrastructure.
Exports remain a key beneficiary. AI-related products — a category that spans semiconductors, power equipment, and consumer electronics — accounted for 20.3% of total exports in 2025. Growth in these categories accelerated to 34.8% year over year during the first five months of 2026 and contributed 6.8 percentage points to overall export growth. At the same time, rising producer prices have begun supporting export revenues and profits, especially across AI supply chains. As a result, we raise our 2026 export growth forecast to 13.0% year over year. Imports could also remain strong at 22.0% year over year, leaving the trade surplus around $1.1 trillion. While protectionist risks remain a concern, we don’t currently see them as large enough to derail overall export growth.
Production appears even more resilient than exports. High-tech industry has increasingly become the main growth driver, with high-tech industrial production contributing more than half of headline industrial production growth despite representing only about 17% of total output. AI-related manufacturing continues to accelerate, with integrated-circuit output rising 25.4% year over year in the first five months of 2026 and industrial-robot production increasing 28.1% year over year.
The benefits of this boom, however, aren’t spreading evenly across the broader economy. Consumer confidence remains subdued, having stayed negative for more than four years. Households continue to save heavily, maintain large excess deposits, and show limited willingness to take on additional borrowing. Meanwhile, fading policy support and earlier stimulus effects contributed to a contraction in retail sales in May, the first decline since Covid.
Property markets tell a similar story. Conditions have improved in some Tier-1 cities, particularly those benefiting from AI-related business activity, but the broader national market remains weak. More generally, AI is creating pockets of strength rather than generating a broad recovery in domestic demand.
Investment trends also highlight the divergence. AI-related investment remains robust, supported by substantial spending from hyperscalers, data centers, and digital infrastructure projects. But investment in many traditional sectors faces mounting headwinds from delayed fiscal deployment, uncertainty linked to geopolitical developments, anti-involution pressures, and squeezed profit margins.
China’s recent reflation story reflects this same uneven pattern. Producer-price improvements have broadened beyond energy, with AI supply chains and construction-related costs also contributing. But profit recovery remains concentrated in upstream and AI-related industries. With domestic demand still weak, we don’t expect significant pass-through from producer inflation to consumer inflation, leaving many downstream businesses facing continued margin pressure.
Against this backdrop, we don’t expect policymakers to pursue large-scale stimulus. In practice, the “AI+” strategy appears to be the dominant policy priority. As long as social stability remains intact and growth stays near target, we believe authorities will focus on facilitating the AI transition rather than trying to eliminate the economy’s underlying divergences.
Therefore, we expect targeted and incremental support rather than broad-based easing. Monetary policy is likely to remain secondary, though we continue to expect a symbolic 10-basis-point rate cut in the year’s second half. Fiscal policy could become more supportive through speedier implementation of existing programs, but we don’t see a significant increase in the budget deficit or government bond issuance as the most likely outcome.
From a market perspective, the K-shaped economy remains evident. Equities tied to AI and the new economy should continue benefiting from concentrated nominal growth, while rates are likely to remain anchored by weaker old-economy dynamics. We also expect the People’s Bank of China to maintain a gradual appreciation bias for the renminbi, supporting its broader internationalization objectives, with USDCNY moving toward 6.7 over the next six to 12 months.
The full report, China Economics: 26H2 Outlook: AI Supercycle Cuts Into the K-Shaped Economy, is available to existing Citi Research clients here.