AI Adoption: Unravelling Its Multi-Directional Impact on the Natural Rate of Interest

AI's Initial Inflationary Dynamics and Productivity

The introduction and widespread adoption of artificial intelligence (AI) across economic sectors could initially generate upward pressure on inflation, primarily through a demand-side mechanism. This effect stems from AI's potential to permanently raise productivity, subsequently boosting incomes for households and firms.

If economic agents rapidly recognise the enduring nature of this productivity improvement and factor future income increases into their current spending decisions, aggregate demand would expand early in the transition phase, leading to price increases.

However, this scenario relies on an assumption that households and firms possess precise knowledge of the nature, scale, and persistence of future productivity changes, which is not a realistic expectation. A more gradual consumption response is a more probable outcome.

This slower adjustment can be explained by 'habit formation' models, where past consumption levels significantly influence current spending benefits, as noted by Philip R. Lane, Member of the Executive Board of the European Central Bank (ECB), in a speech delivered on 6 July 2026.

Furthermore, individuals face considerable uncertainty regarding the specific income implications of the AI transition for their own circumstances, providing an additional incentive to delay consumption adjustments.

It is more likely that households and firms will progressively learn about the income and employment consequences of AI-driven productivity changes over time, adapting their spending in a concurrent and measured manner.

In such an event, any immediate inflationary effect would be considerably reduced, demonstrating the importance of how economic agents process information and adjust expectations.

Labour versus Capital Augmentation and Income Distribution

A critical determinant of AI's macroeconomic influence, including its demand and income distribution effects, is whether the technological advancement primarily augments labour or capital. Economic models frequently represent technology as labour-augmenting, implying that a greater output can be achieved with the same number of workers.

This mechanism typically increases labour income for workers, with the scale of this benefit contingent on their bargaining power and prevailing institutional frameworks. Conversely, if AI predominantly augments capital, the resulting income increases would largely accrue to capital owners rather than to the labour force.

This outcome would intensify the disparity between labour and capital income, contributing to broader wealth inequalities. An expansion in income and wealth discrepancies could, in turn, constrain the degree to which demand grows across all segments of the economy.

Such a limitation on aggregate demand would subsequently temper the inflationary tendencies that might otherwise be associated with AI-driven productivity gains, as outlined by Philip R. Lane of the ECB on 6 July 2026.

The specific nature of AI's augmentation, therefore, holds significant implications for social equity and the overall macroeconomic response to its adoption, influencing the breadth of economic participation and the distribution of benefits from technological progress.

Investment Requirements and Energy Demand Pressures

Integrating artificial intelligence into the economic value chain necessitates substantial investment, particularly in computational infrastructure. Building foundational AI models and implementing AI solutions in various business environments requires significant upfront capital expenditure.

This demand for computing power translates directly into a considerable increase in energy consumption. The expansion of AI-related compute operations will drive up energy demand, and until energy supply systems can adequately adjust to this heightened requirement, upward pressure on energy prices is a probable consequence.

This dynamic is expected to contribute to inflationary pressures during the initial phases of AI adoption, as stated by Philip R. Lane of the ECB in his 6 July 2026 speech.

For instance, research by Burian and Stalla-Bourdillon (2025) suggests that if the electricity demand from AI-driven data centres is met entirely by natural gas, gas prices could rise by approximately 9 per cent in Asia and Europe, and 7 per cent in the United States by 2026, with AI data centres accounting for about 2 percentage points of this increase.

This illustrates a direct channel through which AI deployment can affect broader economic costs and consumer prices, underscoring the interconnectedness of technological advancement, energy markets, and monetary policy considerations during this transition period.

Geographical Distribution and Demand Impact

The geographical distribution of AI activity will materially influence its impact on demand at regional and national levels.

If the development and deployment of AI technology remain largely concentrated in regions such as the United States and China, and if the AI supply chain continues to be heavily centred in Asia, then the increase in investment and energy demand in other regions, such as Europe, might be relatively muted.

In this scenario, Europe could still experience some upward inflation pressure resulting from the effect of increased global demand on commodities and manufactured goods, particularly for products used as inputs into AI production.

Conversely, if there is substantial technological diffusion of AI into Europe, these demand-boosting channels would operate with greater force within the euro area. This effect would be particularly pronounced if technology diffusion requires a degree of local capital investment.

The extent of AI diffusion and the corresponding need for regional capital deployment are critical factors in determining the localised macroeconomic effects, including inflationary dynamics, as highlighted by Philip R. Lane of the ECB on 6 July 2026.

Understanding these geographical patterns is essential for assessing the differential impacts on national economies and for formulating regionally appropriate policy responses to the AI transition.

AI's Influence on the Natural Rate of Interest

The interplay of these macroeconomic propositions regarding AI's effects translates into implications for the natural rate of interest, denoted as R*. This rate is defined as the real interest rate that ensures desired savings align with desired investment, thereby maintaining full employment and stable inflation.

In one direction, a sustained sense of optimism concerning the income and productivity gains anticipated from AI could stimulate investment while simultaneously reducing savings. This combination would exert upward pressure on R*.

Conversely, if households and firms face considerable uncertainty about the future trajectory of AI-induced income paths and the distribution of these income gains across different regions and income groups, the increase in R* would be less pronounced, or potentially absent.

Specifically, uncertainty about potential labour displacement due to AI or about constraints in financing AI-related investment could lead to an increase in precautionary savings. Such an increase in savings, in the face of uncertain income prospects, would act to depress R* rather than elevate it, as discussed by Philip R. Lane of the ECB on 6 July 2026.

The net effect on R* therefore depends on the prevailing sentiment and the degree of certainty regarding AI's long-term economic benefits and distributional consequences, making it a complex variable to forecast.

Trajectory of AI Adoption and R* Over Time

The temporal profile of the natural rate of interest (R*) is also contingent on the pathway of AI technology adoption. One scenario posits that AI will follow a typical S-shaped pattern of diffusion.

In this model, adoption proceeds slowly during its initial stages, accelerates significantly during a phase of widespread implementation, and eventually reaches a plateau as the technology matures. Under this S-shaped trajectory, AI permanently elevates the overall level of productivity within the economy but does not sustain a permanently higher growth rate of productivity.

Consequently, R* would eventually return to the level observed prior to the technological transformation, as the rate of consumption growth would decline once the initial productivity gains abate. An alternative scenario suggests that AI could fundamentally improve the innovation process itself, thereby shifting the economy onto a permanently higher productivity growth rate.

To the extent that productivity growth translates directly into output and consumption growth, in this latter scenario, R* would remain at a permanently elevated level, reflecting the sustained increase in economic dynamism.

The distinction between these two adoption trajectories is crucial for long-term monetary policy planning, as it implies fundamentally different equilibrium real interest rates, as detailed by Philip R. Lane of the ECB on 6 July 2026, building on various economic models.

Investment Volatility and Financial Market Sentiment

Regardless of the specific adoption scenario, the investment rate associated with AI is likely to exhibit considerable volatility. One contributing factor to this volatility arises from potential demand complementarities in implementing innovations, where each innovating sector benefits from other sectors also undertaking innovation.

This creates a positive feedback loop that can lead to rapid surges in investment. Furthermore, financial market sentiment towards AI-related investment may be subject to pronounced waves of optimism and pessimism. This fluctuation in sentiment is understandable given the wide spectrum of views regarding the long-term economic and societal impact of AI.

Indeed, the possibility of multiple equilibria exists, where a transition to a high-capital equilibrium can become self-validating. In such a scenario, optimistic expectations generate a financing feedback loop, attracting capital and driving further investment. During the initial transition to this high-capital equilibrium, investment surges, leading to higher interest rates.

However, as capital becomes abundant and income primarily accrues to high-saving capital owners, the interest rate subsequently falls sharply.

This mechanism, while potentially driving rapid growth, is inherently fragile; a sudden loss of confidence can trigger a self-fulfilling market downturn, as highlighted by Philip R. Lane of the ECB on 6 July 2026, referencing economic theories on speculative growth.

Capital Reallocation and Regional R* Divergence

A distinct scenario emerges if AI production opportunities remain concentrated in specific regions, such as the United States, and if the rate of AI adoption is higher in countries like China compared to Europe.

Under these conditions, Europe could experience a decline in domestic investment, as investors reallocate capital towards the United States and China where AI development and adoption are more advanced.

Even if overseas AI capital can still enhance European productivity through licensing agreements or other intellectual property arrangements, this scenario could generate high incomes in Europe with relatively limited domestic investment. Such a dynamic would entail downward pressure on Europe’s natural rate of interest (R*).

Elements consistent with this scenario are already discernible, including the substantial allocation to US technology stocks within 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 across various middle-tech and high-tech sectors.

This suggests a potential divergence in regional R* trajectories, with capital flows being redirected globally based on perceived AI leadership and adoption rates, as detailed by Philip R. Lane of the ECB on 6 July 2026, drawing on analyses of cross-country AI adoption speeds.

AI as an Amplifier of Cyclical Economic Shocks

Beyond its direct impact on macroeconomic dynamics and monetary policy, AI also holds the potential to amplify other cyclical shocks affecting the economy. Philip R. Lane of the ECB, in his 6 July 2026 speech, outlined three interconnected examples.

Firstly, AI's significant energy intensity means that a persistent upward shock to energy prices could constrain the rate of progress in developing new AI models and curtail the pace of AI adoption.

Secondly, the capital intensity of both AI production and adoption implies that a tightening of financial conditions would negatively affect sectors involved in producing and utilising AI. Thirdly, by offering a substitute for human labour, AI could intensify labour shedding during an economic recession. These channels are not isolated; potential feedback loops exist.

For instance, a prolonged energy shock that alters the economics of AI production and adoption could lead to a repricing of AI-related equity and debt within the financial system. This financial impact could be further amplified if an economic downturn triggered a larger-than-anticipated correction in the labour market, subsequently reducing consumption.

Consequently, a more resilient energy system becomes crucial to mitigate these risks, underscoring that the increasing importance of the energy-intensive AI sector reinforces the rationale for an accelerated transition to an energy system dominated by renewables.

Implications for Monetary Policy and Asian Economies

The various mechanisms through which artificial intelligence may influence macroeconomic dynamics and the monetary policy stance present a complex and uncertain outlook.

Given the many uncertainties surrounding the strength and timing of these diverse channels, a data-dependent approach is best suited for assessing AI's overall impact on the appropriate monetary policy stance, as concluded by Philip R. Lane of the ECB on 6 July 2026. For Asian economies, the implications are particularly salient.

Asia's central role in the global AI supply chain means that increased global demand for AI inputs and components will directly affect regional production and trade flows.

The capital reallocation scenario, where investment shifts towards leading AI hubs, suggests that certain Asian nations could become recipients of significant capital inflows, potentially exerting upward pressure on their domestic natural rates of interest.

Conversely, the substantial energy intensity of AI development and adoption, particularly for large data centres, implies that Asian economies with significant AI investments may face elevated energy costs. For example, Burian and Stalla-Bourdillon (2025) project that AI-driven data centres could contribute to a 9 per cent rise in gas prices in Asia by 2026.

Asian policymakers should closely monitor capital flows into AI-related infrastructure and manufacturing, particularly in nations positioned within the AI supply chain, as these dynamics could exert upward pressure on local interest rates and energy prices through late 2026 and into 2027.

The imperative for an accelerated transition to renewable energy systems is reinforced for the region, aiming to mitigate inflationary impacts and ensure the sustainability of AI-driven economic expansion.