Table of Contents
Dinner speech by Philip R. Lane, Member of the Executive Board of the ECB, at the final conference of the European System of Central Banks Research Network on Challenges for Monetary Policy Transmission in a Changing World (ChaMP)[1]
Rome, July 6, 2026
Let me first congratulate everyone involved in the ChaMP research network on a remarkably successful research program: it has provided many new insights into the transmission of monetary policy and has had a direct impact on our policy discussions in recent years.
In these dinner remarks, I would like to focus on one topic in particular: the impact of artificial intelligence (AI) on monetary policy stance.[2]
A natural benchmark analysis is to view AI as a permanent increase in productivity and income. If households and firms quickly internalize the permanent nature of the productivity shock and factor future income increases into their spending decisions, the emergence and adoption of AI could exert upward pressure on inflation through this demand mechanism early in the transition period.[3]
However, it is hardly realistic to assume that households and firms know exactly the nature, extent and duration of future productivity shocks. A slower consumption response can also be explained if the extent of delayed consumption is an important determinant of the benefits of current consumption, as in “habit formation” models.[4] Consumers also face great individual uncertainty about the impact of the AI transition on their income, which is another reason for slow consumption adjustment.[5] A more plausible assumption is that households and firms simultaneously learn about the impact of productivity shocks on income and employment over time and adjust their spending slowly.[6] In this case, the initial inflationary effect would be greatly weakened.
More generally, the inflationary impact of the AI transition in the context of macroeconomic outcomes arising from varying degrees of incorporation of productivity and income gains into spending decisions will depend on a number of factors.
One factor in determining income, distribution and demand effects is whether the technological boost from AI will be labor- or capital-enhancing. Technology is often portrayed as increasing the workforce: more can be produced with the same number of workers. This effect increases workers’ income, although the amount depends on their bargaining power and institutional factors. On the other hand, if AI increases capital, the increases in income will benefit the owners of capital and not the workers, thereby increasing the inequality of labor and capital income.[7]An increase in income and wealth inequality could limit the extent of demand expansion across all sectors of the economy, thereby dampening the inflationary trends that accompany AI-driven productivity increases.[8]
A second factor is the level of investment required to integrate AI into the economic value chain. Significant computing power is likely to be required here, both for building basic AI models and for implementing AI in business environments. Building the necessary computing infrastructure requires a significant upfront increase in capital expenditures.
A third factor is that the expansion of AI-related computing will bring a significant increase in energy demand and, until energy supplies catch up, put upward pressure on energy prices.[9] These dynamics are likely to increase inflationary pressures during the AI adoption phase.
The geographical distribution of AI activities is likely to be relevant to the impact on demand at a regional or national level. If it turns out that AI activity remains concentrated in the US and China and the AI supply chain remains heavily biased towards Asia, the increase in European investment and energy demand will be relatively muted. In this scenario, Europe would still face some upward pressure on inflation due to the impact of increased global demand for raw materials and goods, particularly products used as inputs for AI production. On the other hand, if there is strong technology diffusion to Europe, these demand-increasing channels will have a stronger impact in the euro area. This is particularly true when technology diffusion can only be achieved with a certain level of local capital investment.
We can translate the competing claims about the degree of anticipation of the macroeconomic impacts of the AI transition into implications for the natural interest rate, defined as the real interest rate that balances desired savings and investments. On the one hand, continued optimism about income and productivity gains from AI would boost investment and reduce savings, putting upward pressure on R*. Conversely, the more uncertain households and companies are about the course of the AI-induced income path and the distribution of income increases across regions and income groups, the smaller the increase in R* would be. In particular, precautionary savings could increase due to uncertainty about workforce relocation or restrictions on financing AI-related investments.
The time profile of R* also depends on the technology introduction process. In one scenario, AI follows the typical S-shaped pattern, where adoption progresses slowly in the early stages, accelerates in a phase of widespread implementation, and finally plateaus as the technology matures. This profile means that AI permanently increases productivity levels, but does not permanently increase the rate of productivity growth.
An alternative scenario, however, is that AI improves the innovation process and thereby shifts the economy to sustained higher productivity growth. To the extent that productivity growth translates one-to-one into production and consumption growth, R* would eventually fall back to the pre-technological transformation level in the former scenario as the consumption growth path becomes lower again after productivity gains fade, while in the latter scenario it would remain permanently at a higher level.[10]
In both scenarios, it is expected that the investment rate will be quite volatile. One source of volatility is that there can be demand complementarities in the implementation of innovations, with each innovative sector benefiting when other sectors also innovate.[11] Financial market sentiment toward AI-related investments may also be subject to waves of optimism and pessimism given the diverse views on the long-term impact of AI. In fact, multiple equilibria can exist, with the transition to a high capital requirement equilibrium being self-confirmed by optimistic expectations, creating a financing feedback loop.[12] When moving to equilibrium with high capital requirements, investments initially increase sharply and the interest rate is high. However, the interest rate then drops sharply because there is an abundance of capital and income flows mainly to owners of capital with high savings. At the same time, this mechanism is inherently fragile: a loss of trust can trigger a self-fulfilling crash.
Finally, if AI production capabilities remain concentrated in the United States and the AI adoption rate in China is higher than in Europe, there is a scenario in which investment in Europe declines and investors reallocate their capital to both the United States and China.[13] In particular, if foreign AI capital can further increase European productivity through licensing agreements, this scenario could still generate high incomes in Europe, albeit with relatively low domestic investment, leading to downward pressure on R* in Europe.
Some elements of this scenario are consistent with the high allocation to US technology stocks in euro area equity portfolios, the high level of European imports of intellectual property products from the United States, and the increasing substitutability between Chinese and European products in a number of medium and high technology sectors.
Given these different mechanisms, the net effect of the AI transition on R* remains uncertain.
So far in this discussion I have focused on the impact of the AI shock on monetary policy. From a broader perspective, it is also important to recognize the potential compounding effect of AI relative to other cyclical shocks that may hit the economy. Let me outline three (possibly related) examples: (a) an energy shock; (b) a financial shock; and (c) a recession shock. The energy intensity of AI means that a sustained upward shock to energy prices could limit progress in building new AI models and also reduce the rate of AI adoption. Due to the capital intensity of AI production and adoption, tightening financial conditions would also have a negative impact on the AI producing and AI using sectors. Finally, AI could exacerbate workforce losses during a recession by providing a replacement for workers.[14]
Clearly there are potential feedback loops between these three channels. For example, a prolonged energy shock that alters the economics of AI production and deployment could also lead to a repricing of AI-related equity and debt in the financial system, which would be exacerbated if it turns out that a downturn in the economy triggers a larger-than-expected correction in the labor market, thereby also reducing consumption. It also follows that a more resilient energy system reduces these risks, so an increasing importance of the energy-intensive AI sector reinforces the logic of an accelerated transition to a renewables-dominated energy system.[15]
Finally, in these remarks, I have provided an overview of the different channels through which AI can influence macroeconomic dynamics and the monetary policy stance. Given the many uncertainties surrounding the strength and timing of the various mechanisms, a data-dependent approach is best suited to assessing the overall impact of AI on the appropriate monetary policy stance. This will be a major challenge for monetary economists and monetary policy makers in the coming years.
https://www.ecb.europa.eu/press/key/date/2026/html/ecb.sp260706~b81aa4e329.en.html
