August 20, 2026, (Inside AI) — Artificial intelligence may supercharge productivity, but that does not guarantee lower inflation, according to new research from the International Monetary Fund's chief economist Silvana Tenreyro.
The paper, published Thursday on the Bank of England's Bank Underground blog, challenges a core assumption among policymakers. Tenreyro co-authored the analysis with BoE economist Jenny Chan and doctoral researcher Ludovica Ambrosino.
At first glance, higher productivity means more output for the same inputs, which should ease price pressures. Federal Reserve Chair Kevin Warsh has publicly hoped AI would let the U.S. economy grow faster without stoking inflation.
But the researchers say the inflation impact of anticipated productivity gains is ambiguous. The timing of spending matters more than the productivity itself.
"Business investment and household spending (can) both move ahead of realised productivity gains, as many argue is happening now with investment in AI infrastructure," the researchers said.
If companies and consumers spend today on the promise of future AI-driven gains, demand can outstrip supply before productivity actually improves. That creates supply crunches, pushes up prices, and may force central banks to keep interest rates higher for longer.
The evidence is already visible. Prices of computer memory and graphics chips have surged over the past year due to data centre demand. Those cost increases are now filtering into phones, laptops, and other consumer electronics.
The research also found that the inflation impact depends on where productivity gains land. Gains in exported goods tend to push up domestic wages and boost demand for supply-constrained services, raising inflation. Productivity improvements in domestically produced services are more likely to lower domestic inflation.
Tenreyro served on the BoE's Monetary Policy Committee from 2017 to 2023. She contributed to the article in her role as a professor at the London School of Economics.
AI's inflation paradox splits central bankers
The findings land amid a heated debate inside central banks. Some policymakers see AI as a deflationary force that will let economies expand without overheating. Others warn that the massive capital spending on data centres and chips is itself inflationary.
The research does not settle that debate. It instead shows that the same technology can produce opposite inflation outcomes depending on how spending and productivity gains are sequenced.
If AI investment arrives before the productivity payoff, inflation rises. If productivity improves first, prices can fall. That sequencing risk is rarely captured in standard economic models.
The paper's publication on a staff blog carries weight precisely because it does not represent the BoE's official view. It signals that the question is open, and that central banks are actively studying AI's second-round effects on wages, services, and supply chains.
What the research leaves unanswered
The analysis focuses on aggregate productivity, but it does not model how quickly AI gains might spread across sectors. It also does not address whether AI could permanently alter the bargaining power of workers, which would change how productivity gains translate into wages and prices.
Those gaps matter for policymakers. If AI primarily boosts productivity in export-oriented tech firms, the inflationary pressure on domestic services could be stronger than headline numbers suggest.
The research was published on the BoE's Bank Underground blog, a forum for staff to share views that do not necessarily reflect the central bank's official position. Reporting by David Milliken; Editing by Susan Fenton.
