Ouyang Xiaohong/Text

A technological revolution is changing the way central banks observe the economy.

From capital investment to industrial applications, and then to financial ecology, AI (artificial intelligence) technology is rapidly iterating and changing one after another. However, the "sensations" of different individuals, enterprises, markets, and even economies are not synchronized.

On October 19, 2026, Pan Gongsheng, the Governor of the People's Bank of China, mentioned at the annual meeting of the Financial Street Forum that there may be a time difference between the demand and supply effects generated by artificial intelligence. In the short term, it will change micro efficiency, stimulate investment, and push up inflation. In the medium and long term, it can improve overall productivity and reduce inflation levels. He also mentioned that the artificial intelligence boom has become one of the main driving forces behind the global stock market's rise, and the global financial market presents high correlation and volatility. He proposed that the next focus should be on three aspects: closely tracking and analyzing the macro impact of artificial intelligence development, doing a good job in financial support for the development of the artificial intelligence industry, and steadily promoting the application of artificial intelligence in the financial field.

Since the beginning of this year, the Federal Reserve, European Central Bank, Bank of Japan, Bank of Canada, Reserve Bank of Australia, and Bank of England have also evaluated the impact of AI on the economy from different perspectives such as investment, prices, employment, interest rates, and financial stability.

The Confusion of Policy Response

The observation focus of central banks varies from country to country. The Bank of Japan and the Reserve Bank of Australia have noted the impact of AI demand on the prices of technology products; The European Central Bank focuses on the distance between investment expansion, technology application, and productivity; The Bank of Canada studies policy trade-offs under changes in employment structure; The Federal Reserve discusses potential growth and labor, while the Bank of England includes AI financing in financial stability monitoring. These studies are focused on the fact that a unified conclusion has not yet been reached, and it does not mean that countries have adjusted their policies accordingly.

There is a phenomenon worth noting: AI is causing a "multiple speed mismatch" in the economy (time difference, speed difference, and expectation difference). The continuous breakthroughs in AI technology capabilities have led to the formation of real demand for capital expenditures, and financial markets have included future returns in advance. Productivity improvement, employment adjustment, and institutional adaptation are unfolding along different timelines.

Here, there are both familiar technological diffusion delays in economics and cognitive and decision-making pressures brought about by the accelerated evolution of new technologies. Investment growth does not necessarily mean a synchronous increase in potential supply capacity, and market pricing cannot replace real cash returns.

For the central bank, the challenge lies in determining what is happening in the economy at present, distinguishing which changes belong to short-term demand, which may lead to sustained supply improvement, and how the unrealized future will affect today's financial conditions in advance.

Let's start from the starting point of AI investment. From chips and servers to data centers, power grids, and cooling systems, enterprises procure equipment, organize construction, arrange energy and financing. These expenditures quickly generate orders and current demand, but whether the equipment can be fully utilized and how much new revenue can be created after completion still needs to be gradually tested.

On July 28, 2026, the Bank for International Settlements (BIS) published "AI and the Global Economy: Implications for Central Banks", which pointed out that AI is driving large-scale, increasingly debt financed investment expansion and affecting the global economy through trade, financial markets, and wealth effects. Although the potential productivity benefits of AI are considerable, there is still uncertainty and uneven distribution among different countries and industries.

The Bank of Japan has included AI related demand in its official forecast for July's Economic and Price Outlook, which supports Japan's economic growth and drives up prices of commodities such as semiconductors. Affected by factors such as energy prices, yen depreciation, and AI related demand, the Bank of Japan expects that the year-on-year increase in consumer prices in the second half of the 2026 fiscal year may be significantly higher than 2%.

This does not mean that AI has become the main source of inflation in Japan, but it indicates that a technology that is expected to reduce long-term costs is having an impact on real prices.

The Reserve Bank of Australia stated in its monetary policy statement on September 29th that AI related demand is driving up global technology product prices rapidly.

The scale of capital investment in Europe is also visible. On October 13th, Philip Lane, a member of the Executive Committee of the European Central Bank, revealed that Eurozone businesses plan to allocate approximately 10% of their investments in AI by 2026, with AI related loans accounting for about a quarter of recent corporate credit growth.

Ryan previously pointed out that if households and businesses incorporate future productivity gains into their consumption and investment decisions in advance, AI may create inflationary pressures through demand channels in the early stages of transformation.

Why does a technology that may reduce production costs in the future increase the cost of some economic activities in the early stages of development?

In her 2002 book "Technological Revolution and Financial Capital," technical economist Carlotta Perez distinguished major technological changes into the "installation period" and the "deployment period": capital takes the lead in promoting infrastructure construction and industrial expansion, while technology can only release productivity benefits more widely through enterprise organization, business models, and institutional adjustments.

There have been varying degrees of time lag between capital formation and technology popularization, from railways and electricity to information technology. Perez's theory provides historical coordinates for observing today's AI, but it cannot be concluded that it will inevitably repeat past financial cycles.

What the central bank needs to distinguish is whether the current growth mainly comes from construction activities or whether the conditions for sustained improvement in supply capacity have already been formed? This will affect its judgment on the nature and persistence of inflation.

When will productivity be realized

If changes in orders, investments, and prices can be observed quickly, improvements in productivity often take longer.

On October 13th, Ryan revealed the results of a survey conducted by the European Central Bank, which showed that 64% of surveyed Eurozone companies were already experimenting or using AI to a moderate extent, with only 7% considering their level of application to be "significant"; 52% of surveyed workers reported using AI in their work, compared to 26% in 2024.

Users have reported significant time savings, but the benefits are mainly focused on specific tasks such as programming. The European Central Bank has not yet observed a significant and identifiable impact of AI on overall productivity.

A tool has entered the work schedule of more and more people, why hasn't it left a noticeable mark in macro statistics?

The March paper by Bank of Canada researchers Joshua Brault, Maryam Haghighi, and Jing Yang referred to this phenomenon as the "AI productivity paradox". Some micro empirical studies have shown that AI can increase productivity for specific tasks by 15% -60%, but the related benefits have not yet been fully reflected in macro productivity statistics.

These numbers come from different tasks and samples, and cannot directly calculate the efficiency improvement of the entire economy.

Classic research provides another layer of explanation for this. In 2021, a study on the "productivity J-curve" published by economists Eric Brynjolfsson, Daniel Rock, and Chad Safran showed that after the use of general-purpose technologies, companies often need to synchronously adjust business processes, train employees, transform organizations, and invest a large amount of intangible assets that are difficult to fully measure.

These initial investments may not immediately bring visible output growth. With the gradual maturity of new technologies and supporting organizations, productivity benefits may further manifest.

Therefore, the increase in technology utilization and macro productivity growth are not significant and can be simultaneously established. However, statistical lag cannot guarantee a significant increase in productivity in the future.

Technological progress will also change labor demand.

The "dual speed economy" model established by the same research team of the Bank of Canada in July this year distinguishes the impact of AI into assisting workers in improving efficiency and automating replacement tasks. Research has found that the impact on different industries may vary significantly. If the central bank attempts to stabilize employment in affected industries by expanding aggregate demand, it may also increase inflationary pressure on other industries.

This is a policy trade-off revealed by theoretical models, not a realistic judgment that the Bank of Canada has already experienced AI unemployment.

In August, Federal Reserve Chairman Kevin Walsh raised a series of undecided questions during his speech at the Jackson Hole Central Bank Annual Meeting: When will AI bring sustained economic productivity gains? Are computing power and labor complementary or competitive? Who ultimately receives the return on investment in technology?

These issues are related to potential economic growth and how the central bank evaluates employment, wages, and inflation.

Ultimately, the long-term economic value of AI still depends on its practical applications. The improvement of model capability can only be further transformed into enterprise process improvement, production efficiency enhancement, and new products and services, in order to enter the national economic statistics more widely.

The answer to interest rates varies

The changes in productivity and employment are still unfolding, but the financial market is forming prices for the future every day.

The Financial Stability Report released by the Bank of England in July pointed out that the investment demand of some AI companies in 2025 has exceeded the scale that their operating cash flow can support. External financing will further accelerate in the first half of 2026, involving various channels such as corporate bonds, bank loans, private equity credit, and structured financing.

The market forecast cited in the report shows that the capital expenditure expectations for large AI technology companies in 2028 have been revised upwards from less than $600 billion by the end of 2025 to over $1 trillion.

The financing structure is changing the funding constraints of different enterprises in the AI industry chain.

Data center buildings and supporting facilities can be used for a long time, but internal chips and servers face faster technological iterations. Whether a long-term financing can be continuously repaid by the project's cash flow depends on equipment updates, customer contracts, utilization rates, and future financing conditions.

The Bank of England is therefore concerned about how risks may be transmitted as AI construction shifts from internal corporate financing to extensive credit markets. But the report also clearly states that there is currently insufficient evidence to suggest that AI financing has clearly squeezed the ability of other companies or governments to obtain funding.

In October, the International Monetary Fund's Global Financial Stability Report further warned that if AI investment and profit expectations fail to materialize, asset price adjustments may amplify market volatility through leverage, concentrated holdings, and credit linkages.

This is a risk scenario that needs to be monitored and does not mean that the technology investment cycle has reversed.

More subtle changes occur in interest rates. On October 13th, Ryan pointed out that the expansion of global AI investment and expectations for future productivity improvements may be having an upward impact on global long-term interest rates. The scale of AI investment in Europe is relatively limited, but it is also influenced by changes in the international bond market.

However, long-term interest rates are simultaneously influenced by factors such as inflation, fiscal financing, savings, risk premiums, and expectations of monetary policy. The increase in AI investment does not necessarily mean that natural interest rates will rise.

If AI continues to improve productivity, it may expand profitable investment opportunities, increase income, savings, and improve corporate debt paying ability. Its net impact on long-term equilibrium real interest rates still requires data verification.

Perez's distinction between financial capital and productive capital still has enlightening significance: the capital market can price technological prospects in advance, while enterprises must complete construction, sales, and collection over a longer production cycle.

Technology can continue to break through, but asset prices may not necessarily rise synchronously. The changes in financing costs will also have a reverse impact on which projects can be sustainably constructed and which innovations can receive capital support.

This makes the relationship between AI and monetary policy more complex. The central bank needs to observe the price pressure that investment demand may bring, evaluate the impact of technology diffusion on potential growth, and also pay attention to credit expansion, financing structure, and new risks arising from the use of AI by financial institutions themselves.

The policy direction proposed by Pan Gongsheng also has practical significance. Pan Gongsheng pointed out that in the early stage, the People's Bank of China, together with relevant departments, studied the work ideas and directions for promoting the application of artificial intelligence in the financial field from the aspects of improving infrastructure, enhancing application capabilities, expanding application scenarios, and preventing and controlling security risks, and formulated and issued relevant policy documents on "artificial intelligence+finance". The People's Bank of China will continue to explore the application of artificial intelligence in assisting in monetary policy situation analysis, financial risk identification, as well as payment, credit reporting, anti money laundering regulatory services, and other fields; At the same time, guide financial institutions to prevent risks such as data security and model application.

Major central banks around the world are paying attention to, researching, and evaluating AI, facing several economic clocks with different speeds: investment has occurred, technology applications are spreading, productivity gains are yet to be verified, and interest rates and asset prices are forming every day.

Monetary policy needs to make judgments on current growth, prices, and financial stability before they are realized in the future.