Artificial intelligence is widely expected to make workers and businesses more productive, yet the enormous investment required to build the technology could push prices higher before those efficiency gains arrive. Research involving International Monetary Fund chief economist Silvana Tenreyro argues that expectations of future productivity can trigger investment and spending immediately, creating inflationary pressure while the economy is still waiting for the extra productive capacity to materialize.
That creates an unusual economic sequence. Companies spend heavily on data centers, processors, electricity infrastructure and skilled employees today because they expect AI to reduce costs tomorrow. Research by Tenreyro and her co-authors suggests this front-loaded demand can strain available resources and lift prices, even when the technology eventually allows the economy to produce more efficiently.
Why the AI Buildout Can Raise Costs First
The clearest pressure comes from physical infrastructure. AI systems require large data centers filled with advanced computing equipment, cooling systems and networking hardware. The International Energy Agency reports that global investment in data centers reached roughly $500 billion in 2024, nearly twice its 2022 level. The construction boom also creates demand for generators, steel, cooling equipment, cables and other industrial products.
Semiconductors provide another example. Strong demand for graphics processors, memory and related components can run ahead of manufacturing capacity. Tenreyro’s research, co-authored with Bank of England economist Jenny Chan and researcher Ludovica Ambrosino, points to rising prices for computer components as evidence of how an investment surge can create bottlenecks before productivity improves.
Electricity adds another layer. The International Energy Agency estimates that data centers consumed about 415 terawatt-hours of electricity worldwide in 2024. Consumption could more than double to around 945 TWh by 2030, with AI accounting for much of the increase. Data centers are expected to generate nearly half of the growth in U.S. electricity demand through the end of the decade.
Investment Today, Productivity Tomorrow
The inflation question therefore depends heavily on timing. Businesses may need to hire engineers, electricians, construction workers and other specialists while competing for equipment and energy. Higher demand can support wages and supplier prices before AI has meaningfully reduced the amount of labor, time or capital required to produce goods and services.
The longer-term picture may look different. Research published by the IMF argues that AI has significant potential to lift global growth, particularly if businesses learn to integrate it effectively. Investment in U.S. information-processing equipment and software was already growing rapidly during 2025, reflecting expectations that automation and better decision-making could eventually increase output per worker.
If companies can produce more with the same resources, unit costs may fall. Faster productivity growth could allow wages and profits to increase without creating the same degree of price pressure. That is the traditional argument for viewing technological progress as disinflationary.
What Does This Mean for Interest Rates?
For central banks, the transition creates a difficult problem. Policymakers cannot base interest-rate decisions solely on AI’s expected future benefits if today’s investment boom is increasing demand faster than supply can respond. Minutes from the Federal Reserve’s July 2026 meeting showed policymakers remained concerned about persistent inflation and the possibility that higher rates could become necessary if price pressures failed to ease.
Consumers could feel the consequences through borrowing costs. If AI-related spending contributes to persistent inflation, interest rates may remain elevated for longer, affecting mortgages, auto loans and credit cards. Investors face a similar tradeoff. AI investment may support corporate earnings and economic growth, while higher bond yields can reduce the present value investors assign to future profits, particularly for growth-oriented technology companies.
Where Productivity Gains Occur May Matter Most
The final effect may depend on which parts of the economy become more productive. Tenreyro and her co-authors argue that productivity improvements in domestic services could help reduce local price pressure. Gains concentrated in export industries may work differently, potentially raising incomes and demand for services at home.
That leaves AI with two competing economic identities. It is a massive investment cycle consuming chips, electricity, capital and skilled labor today, and a technology that could make the economy substantially more efficient tomorrow. Whether AI ultimately proves inflationary or disinflationary may therefore depend less on its capabilities alone and more on how quickly productivity arrives, where those gains appear, and whether supply can expand fast enough to keep pace.
