China’s AI frenzy

AI optimism has revived short-term confidence in China’s economy.

“China’s recent private business symposium with the authorities on advanced technology sectors signalled heightened policy support for AI, further speeding up adoption.”

The country’s equity market has rallied strongly since the middle of January and the surprising performance can be attributed to several developments.

Primary among these is China’s progression in Large Language Models (LLMs) that has triggered a global repricing of technology.

At the same time President Xi Jinping has met business leaders, mainly from the tech sector, signalling the government’s determination to revive investor confidence.

And, while the US imposed a 10 per cent additional tariff on Chinese goods, we estimate that the total impact of this additional tariff is only 0.5 per cent of the Gross Domestic Products. The impact of another 10 per cent will likely be marginal.

Spread across several years China will likely target 5 per cent growth in 2025, evidenced by the average provincial GDP targets of 5.3 per cent.

Helping people spend

The authorities will extend supportive measures on domestic consumption and investment as a response to increased external uncertainty.

Historically, China’s stock market is a good indicator of economic confidence, as measured by monthly Purchasing Managers’ Index.

The rapid adoption of AI coupled with a growth-oriented policy stance will likely increase the capex of both state-owned enterprises and the private sector.

This will extend the positive momentum built from the fourth quarter of 2024 (GDP was at 5.4 per cent year on year and 1.6 per cent quarter on quarter).

AI adoption in China

Researchers have attempted to project the macroeconomics of AI.

By extending the company-level evidence of AI adoption and productivity gains, a 2024 OECD study postulated that 23–50 per cent of US companies will adopt AI over the next 10 years.

This study estimated AI could lift long-term Total Factor Productivity (TFP) in the US by an annual rate of 0.24–0.53 per cent.

The growth trajectory is likely to resemble the discovery and application of electricity in the 19th century, computers in the post-war era and the Internet in the 1990s.

China’s adoption of AI appears to be faster than that of the US.

A 2024 SAS survey of 1600 companies indicated that China has an adoption rate of 83 per cent compared to 65 per cent in the US.

China’s recent private business symposium with the authorities on advanced technology sectors signalled heightened policy support for AI, further speeding up adoption.

US productivity currently outpaces that of China’s. Faster adoption of AI in China could narrow this gap and see China’s TFP surpass 0.5 by 2030. This would also enable China to catch up with other East Asian countries such as Japan and South Korea.

A study by the UN’s World Intellectual Property Organization shows the number of Gen AI patent applications by Chinese companies in 2014–23 represents 70.3 per cent of the world’s total, compared with 11.5 per cent by US companies.

Our local observations indicate a fast and high diffusion of AI applications in many segments of China’s economy, from manufacturing to services.

Predicting China’s AI economic boost

AI will help China address labour market constraints from demographic challenges.

Wang et al (2023) estimated that 54 per cent of tasks in China’s job market could be replaced by AI, offering a large efficiency gain.

The job displacement is worrying, but the income effect can ultimately offset the substitution effect in the long run.

Earlier, PWC (2018) projected that AI will deliver a net job creation of 93 million in China.

The structural benefit of adopting new technology and improved productivity is apparent in the long run.

Our core view requires China to significantly improve TFP to break the middle-income trap.

AI adoption is an opportunity. However, the tech-driven regime is probabilistic, not deterministic.

While the prospect of TFP improvement is compelling, one or two technological breakthroughs is not sufficient for us to revise China’s GDP forecast upward.

We will watch closely how AI adoption translates into hard economic variables.

The “AI shock”

A technological shock can lift potential growth. But the cyclical performance of any economy is measured by the difference between actual demand and potential output.

The gap could be negative with excess capacity and increased joblessness, especially during the phase of job displacement.

One major topic in AI discussion is the Baumol effect, which means the strong demand in tech-related segments could spill over to the non-AI sectors, eventually leading to high wages and lower growth.

However, we do not think this effect will kick in given the overcapacity issue in China.

The fast pace of AI adoption in the last few years has not saved the Chinese economy from falling into a deflationary trap.

Despite being structurally positive, initial cyclical economic indicators tend to head south because labour market transition is slow.

Any displacement of workers initially means loss of household income, which translates into a potential decline in private consumption.

Demand will catch up with potential growth only when the displaced capacity, including workers, integrate into the new regime and share the income growth.

AI as a technological shock is akin to emerging markets joining global supply chains a few decades ago.

To advanced economies, China’s worker pool and India’s shared services seemingly presented a ‘human version of an AI’ decades ago. The effect was an overall decline in price levels in the advanced states.

Likewise, the AI shock will also lower price levels of the old economy in China.

Investment yield and interest rates

Structurally, AI is positive to the Chinese interest rate outlook.

A well-known conclusion from the secular stagnation in the past 40 years is that real interest rates follow potential growth, which is determined by productivity and other structural factors.

Real rates will rebound if productivity improves.

However, the process will not be linear.

As China continues its transition to an innovation-driven growth model, the drag from the old economy will continue before the rise of new sectors.

Both property and infrastructure investment have seen diminishing returns in the past few years, which translates into downward pressure on interest rate.

Recent AI breakthroughs by Chinese tech companies will reinforce the People’s Bank of China’s cautiousness in monetary policy easing.

The authorities are mindful of the drawback from the low-rate regime.

In fact, they have warned about zombie companies, debts and overcapacity due to low rates several times in the past.

After the interest rate cut last September, the authorities have not lowered the policy interest rate and required reserve ratio (RRR) even with relatively tight liquidity conditions prior to the Lunar New Year holiday.

Instead, they are using non-conventional methods to support the capital market instead of interest rate cuts.

The bottom line

The Chinese equity market has been optimistic about the technology sector in the last few weeks.

And as a technological breakthrough, AI could potentially lift China’s TFP and lift the growth profile in the long run.

At present, ANZ Research sees it as too early to equate these impressive developments in the LLM application with a macroeconomic event or consider them a cyclical shock.

But what can be certain is that the Chinese authorities have an eye to turning this short-term confidence boost into something much more sustained.

Raymond Yeung is Chief Economist Greater China at ANZ, Vicky Xiao Zhou is an Economist at ANZ and Zhaopeng Xing is Senior China Strategist at ANZ