LONDON — Artificial intelligence is widely expected to transform economies by making workers and businesses more productive. But a new analysis involving International Monetary Fund chief economist Silvana Tenreyro warns that higher productivity does not automatically mean lower inflation — particularly when businesses and households spend heavily on AI before its promised economic gains fully materialize.
The research, co-authored by Tenreyro, Bank of England economist Jenny Chan and doctoral researcher Ludovica Ambrosino, argues that the inflationary impact of productivity improvements depends on timing, demand, supply and monetary policy.
That finding challenges a popular assumption surrounding the global AI boom: that technologies capable of producing more output with fewer resources will eventually reduce costs and ease pressure on prices.
The AI problem starts before the productivity gains arrive
The researchers’ key warning is about anticipation.
Companies may invest in data centres, chips, software and other AI infrastructure because they expect major productivity gains in the future. Consumers and businesses may also increase spending based on expectations of stronger future incomes.
But if that demand arrives before the additional productive capacity does, the economy can experience supply bottlenecks.
That imbalance can push prices higher rather than lower — potentially forcing central banks to keep interest rates higher to prevent inflation from becoming entrenched.
The phenomenon is already visible in parts of the technology supply chain. Reuters reported that prices for computer memory and graphics chips have risen sharply amid strong demand from data centres, contributing to higher prices for products such as smartphones, laptops and other electronics.
Productivity is not automatically deflationary
The economics behind the warning is more complicated than the simple equation of higher productivity = lower prices.
When productivity improves, companies can produce more with the same amount of labour and capital. That can reduce production costs and increase the economy’s capacity to supply goods and services.
But productivity can also increase people’s expected incomes and encourage households and companies to spend more.
If demand expands faster than supply, inflationary pressure can emerge.
Tenreyro and her co-authors distinguish between a temporary increase in productivity and a persistent improvement in productivity growth. A temporary productivity boost can lower marginal costs and put downward pressure on prices. But sustained productivity growth can raise expected permanent income, encouraging investment and consumption.
The result is that the effect on inflation is ambiguous rather than guaranteed.
Where the productivity gains happen also matters
The research points to another important factor: which part of the economy benefits from productivity improvements.
Productivity gains in domestically produced services are more likely to ease domestic inflation because they increase the economy’s ability to supply services.
But productivity gains concentrated in export industries can have a different effect.
Stronger export performance can increase wages and incomes, which may then raise demand for domestic services whose supply cannot expand as quickly. That can create additional inflationary pressure.
In other words, the same technological revolution can have very different inflation effects depending on where the gains occur and how quickly the rest of the economy adjusts.
The warning comes as central banks reassess AI’s economic impact
The debate is becoming increasingly important for monetary policymakers.
US Federal Reserve Chair Kevin Warsh has argued that AI could eventually allow the US economy to grow faster without generating equivalent inflation pressure.
But other central-bank research has highlighted the opposite short-term risk.
The Bank of Japan recently warned that strong global AI-related demand could create persistent inflationary pressure even though AI is expected to improve productivity and reduce prices over the longer term.
The European Central Bank has likewise examined how optimism over AI-driven income and productivity gains can stimulate investment and alter the balance between saving and spending, potentially affecting interest rates and the broader economy.
Meanwhile, the Bank for International Settlements has warned that the AI investment boom could create new challenges for central banks because investment, asset prices, productivity and inflation may move in different directions.
The bigger question: Can AI deliver its gains fast enough?
This is where the AI inflation debate could become especially important.
If AI productivity improvements arrive rapidly enough to expand supply alongside demand, the technology could ultimately become a powerful disinflationary force.
But if companies pour enormous amounts of money into AI infrastructure first — while chips, electricity, data centres and other inputs remain constrained — inflation could rise before the productivity benefits appear.
That creates a difficult policy problem.
Central banks cannot simply assume that future technological gains will automatically solve today’s inflation. They must respond to the economic conditions that actually exist while trying to estimate how quickly AI will change productivity in the future.
The IMF’s latest global outlook underscores that inflation remains a major issue: the Fund expects global headline inflation to rise from 4.1% in 2025 to 4.7% in 2026, before easing to 3.9% in 2027.
What this means for the global economy
The AI revolution could still deliver enormous long-term productivity gains. But the new research suggests that the path from technological progress to cheaper goods and services is far from automatic.
The crucial question is not simply whether AI makes economies more productive.
It is when those productivity gains arrive, how consumers and companies respond, where the gains are concentrated, and whether supply can keep up with the surge in demand.
That means the technology that promises to make economies more efficient could, paradoxically, create more inflation pressure before it delivers its biggest productivity payoff.
And for central banks already struggling to balance economic growth against persistent inflation, that could make the AI boom one of the most complicated policy challenges of the decade.
Leave a Reply