Why Isn't AI Saving China's Economy?

The Chip Wall: Hardware Constraints Are a Binding Limit

China's AI ambitions are literally hitting a silicon wall. For over half a decade, the United States has imposed escalating semiconductor export controls designed specifically to slow China's chip industry and preserve US leadership in the computing capabilities that undergird AI advances . These controls have restricted access to advanced GPUs and the equipment needed to manufacture them domestically. The impact is visible: despite heavy stockpiling of Nvidia H800 and H20 GPUs in 2024, these high-performance chips are difficult to find in Chinese cloud services .

In response, Beijing mandated in late 2025 that state-funded data-center initiatives use only domestically produced AI chips . However, Chinese chipmakers are struggling with manufacturing bottlenecks. SMIC, China's leading foundry, remains stuck at 7nm process technology due to US and allied export controls, making it hard to produce competitive AI chips at scale . This forces the AI sector to optimize for efficiency with constrained hardware rather than scaling up raw compute to drive economic spillovers.

The Numbers: AI Adds a Fraction, Property Subtracts a Chunk

Quantitative evidence shows that even a booming AI sector cannot fill the hole left by real estate. One analysis from Caixin found that in China, AI contributed only about ~0.3% to GDP, far smaller than the drag imposed by the property sector . A separate analysis by the Rhodium Group covering 2023 to 2025 found that emerging sectors like AI, robotics, and EVs contributed a mere 0.8 percentage points to economic performance, while traditional industries, particularly real estate, experienced a combined decline of 6 percentage points .

The property crisis is simply a different type of economic problem than the one AI solves. A real estate downturn decimates household wealth—since property accounts for a huge share of Chinese household assets—destroys confidence, and freezes construction activity . Each further 1% decline in housing prices can reduce private consumption by roughly 0.2% of GDP . AI is a productivity and investment story that works through capital markets and tech supply chains. It cannot quickly revive housing demand, restore household balance sheets, or boost consumer confidence at the scale needed to drive a recovery.

Economic Bifurcation: A Boom for the Few, Not the Many

Instead of acting as a unifying economic force, the current AI boom is creating a "K-shaped" divergence in China's economy . The economy is increasingly splitting into two different stories: a globally competitive, high-tech segment advancing rapidly in AI and exports, and a domestic segment weighed down by falling home prices, weak consumption, and employment uncertainty .

Prominent voices have highlighted this risk. Lu Ting, chief China economist at Nomura, has warned that AI can further widen China’s economic bifurcation rather than act as a panacea . The benefits of AI are accruing first to firms, regions, and workers who can access capital, high-end compute, and engineering talent—concentrated in a handful of tech hubs—while small business owners elsewhere struggle as their clientele cuts back on spending . Even the World Bank has noted that technological change, including AI, is displacing low-skilled jobs while increasing demand for high-skilled talent, disproportionately affecting informal and migrant workers .

The U.S. Comparison Is Also Overstated

It is also important to note that the narrative of AI "driving half of the US economy" is not fully supported by evidence. While AI-related capital expenditures provided a significant boost to US GDP growth in 2025, their net contribution is smaller once the high import content of AI hardware is accounted for . Research from the St. Louis Fed and other economists indicates that after adjusting for imports of computers, semiconductors, and other AI-related equipment, the net average contribution diminishes significantly, translating to approximately 20%-25% of real GDP growth rather than the higher headline figures . Even in the US, where investment is over a dozen times larger than China's, the macro payoff is meaningful but not dominant—and it relies on a domestic chip design and capital market ecosystem that China cannot currently replicate.

Long-run forecasts reinforce that AI is a gradual productivity play, not a cyclical rescue tool. Penn Wharton estimates that AI will raise US GDP by just 1.5% by 2035, with more substantial gains of nearly 3% by 2055 and 3.7% by 2075 . Those timelines are irrelevant for an economy that needs demand support today. As one analyst noted, the short-term outlook for China remains overshadowed by the property crisis, a lack of dynamism in the labor market, and low household confidence, making AI a powerful but poorly matched tool for immediate macro repair .