Will Artificial Intelligence (AI) bring down inflation or make it worse? The question has become a hot topic among economists, central bankers and investors. The Fed's new Chair, Kevin Warsh, believes AI "will be a significant disinflationary force" [1], boosting productivity and reducing costs across the economy.
Yet the AI boom is already driving a surge in demand for data centres, energy and computer chips, pushing up prices for a range of goods and services. In this month's article, we explore why AI could eventually prove disinflationary, but why its immediate impact is likely to increase inflationary pressures.
The productivity promise
The case for AI bearing down on inflation rests on several channels. By replacing relatively expensive labour with increasingly capable software and automation, AI has the potential to reduce firms' operating costs. By automating routine tasks and improving efficiency, AI could ultimately allow workers to focus on higher-value activities and become more innovative.
In turn, this holds out the prospect for a sustained improvement in productivity growth, which could lower unit labour costs and ease pressure on prices. If these gains prove widespread and persistent, the economy could produce more goods and services with the same amount of labour and capital, increasing growth while reducing inflationary pressure.
Furthermore, if AI leads to job losses and the unemployment rate rises, greater labour-market slack could result in slower wage growth, which could further restrain inflation and aggregate demand. Indeed, the mere risk of AI-related job losses might result in workers reining in their wage demands, which could in turn be disinflationary.
While these trends are theoretically possible, there is currently very little empirical evidence to suggest that AI productivity gains are resulting in lower inflation at the economy-wide level. Surveys of company executives suggest that AI is improving productivity at some US firms, but as yet, a break-out in overall US productivity growth is elusive.
One study of 734 executives by economists at the Richmond Fed concluded that AI-attributed productivity growth was about 0.6% in 2025, with the strongest effects concentrated in high-skill services and finance, where AI-driven productivity gains exceeded 2%[2].
However, these gains have yet to show up in an acceleration in economy-wide productivity growth. Overall US labour productivity is running at an annualised pace of 2.1% during the current business cycle, which is the same as the long-term rate of 2.1% since the first quarter of 1947[3].
AI and the labour market
Similarly, the impact of AI on the labour market and aggregate wage setting behaviour is uncertain. AI is increasingly being cited as the major reason for job cuts in the US, with a recent survey suggesting that AI accounted for 40% (or 38,579) of all job cuts announced last month, with tech companies leading AI-related redundancies[4].
More broadly, research from the Federal Reserve Bank of Dallas shows that employment in the 10% of sectors most exposed to AI (such as computer systems design, law etc.) is falling slightly, not least as young workers find it increasingly difficult to find work in these areas. However, the economists at the Dallas Fed found that during the three years following ChatGPT's release in November 2022, wages in the 10% of industries most exposed to AI rose at a faster rate, by 8.5% compared with 7.5% across the economy as a whole[5]. This occurred because AI is cutting entry-level jobs through automation, while simultaneously boosting the value (and pay) of the highly experienced workers who remain.
At present, AI does not appear to be depressing economy-wide wage growth, but it is altering the distribution of that growth. AI is showing signs of suppressing the earning power of entry-level workers and those performing routine cognitive tasks, while massively amplifying the earnings of experienced professionals and tech-literate specialists.
As regards employment, at the economy-wide level, the US continues to create jobs on a net basis, with the pace of hiring actually picking up in recent months[6]. Indeed, the number of job vacancies has recently risen above the number of unemployed workers in the US for the first time since June 2025[7], suggesting the US labour market is tightening.
In terms of the actual inflation data in the US, we are yet to see any downward pressure on prices resulting from AI. Headline inflation in the US accelerated to a 3-year high of 4.2% in May[8], largely as a result of rising energy prices due to the disruption in the Strait of Hormuz. However, the increase was not confined to energy alone. Excluding food and energy prices, so-called core inflation also rose, reaching 2.9%[9], its highest level in eight months.
Fed Chair Warsh has indicated that he favours so-called trimmed mean measures of inflation (see our June 2026 article[10]). However, even on this measure, underlying inflation picked up in May, rising to 2.9%[11], its highest level in five months.
The cost of the AI build-out
Indeed, rather than exerting downward pressure on inflation, the ongoing surge in AI infrastructure investment appears to be driving up the prices of certain goods and services. Bank of America (BoA) estimate that investment in AI infrastructure (data centres etc.) will rise 67% to US$800bn in 2026, and climb further to US$1 trillion in 2027[12].
This investment boom is driving a surge in demand for the semiconductors, memory chips, networking equipment and other hardware needed to train and run AI models. In many cases, supply has struggled to keep pace, leading to sharp price increases. The price of Dynamic Random-Access Memory (DRAM), which stores data in capacitors and transistors, rose 95% in the first quarter alone[13], as strong demand from AI data centres and a shift by manufacturers towards higher-end AI memory products tightened supply. Looking ahead, research firm Gartner forecasts that combined DRAM and solid-state drive (SSD) prices will be around 130% higher by the end of 2026 than they were in 2025[14].
These wholesale price increases are only just starting to feed through into consumer prices. Having fallen for most of the last 20 years (and exerted downward pressure on inflation) the computers and peripherals sub-index of the US CPI rose 1.3% year on year in May[15]. Further increases appear likely as manufacturers pass higher input costs on to customers.
Goldman Sachs expects average selling prices for computers and non-Apple smartphones to rise by roughly 10% this year. As a result, Goldman expects higher electronics and software prices to boost annual inflation on the Fed’s preferred core PCE measure by 0.3 percentage points over the next year[16]. Given that semiconductors are embedded in a vast range of products, from household appliances to motor vehicles, the broader inflationary impact could prove even greater.
In addition to rising hardware costs, some companies are also increasing software prices. In January, Microsoft raised the price of its Microsoft 365 subscriptions by between 30% and 43% (depending on the plan), marking its first increase in 12 years. The company said the move reflected "the extensive subscription benefits that we've added over the past 12 years" and would help it "deliver new innovations for years to come"[17]. Adobe has also announced price increases for some of its software products.
While the addition of AI-powered features arguably means customers are now receiving enhanced products and services, the price increases also suggest that firms are seeking to recoup at least part of the substantial investment required to develop AI capabilities and expand supporting infrastructure. As a result, some of the costs associated with the AI boom may ultimately be passed on to consumers through higher software prices.
Powering the AI revolution
The AI investment boom also appears to be boosting electricity prices in the US. AI data centres use massive amounts of electricity to continuously run and cool the high-powered computer hardware needed to crunch vast amounts of data.
Research from the Dallas Fed notes that after two decades of broadly stagnant U.S. electricity demand, the rapid growth of generative AI workloads has led to a resurgence in power consumption. The researchers found that existing data centres have already increased wholesale electricity prices by between 2% and 6% on average nationwide, with substantially larger effects in regions hosting major data centres. Going forward, their modelling suggests that if planned data centre construction proceeds and utilisation rates remain high, wholesale electricity prices could rise by around 50% by 2028.
The electrification associated with AI is also contributing to higher prices for raw materials such as copper. Copper is a critical component in power generation, transmission and distribution networks, as well as in the servers, cooling systems and other infrastructure used by data centres. The expansion of electricity grids and construction of new AI-related infrastructure is therefore adding to copper demand at a time when supply growth remains constrained.
Copper prices have risen to record highs in recent weeks[18], and growing demand from AI-related infrastructure projects could place further upward pressure on prices in the years ahead. As copper is used extensively in electronics, vehicles and industrial equipment, higher prices are likely to raise production costs across multiple sectors, ultimately putting upward pressure on consumer prices.
The wealth effect
In addition to boosting investment demand, AI may also be supporting consumer spending as the recent AI-related rise in equity prices increases household wealth and encourages greater expenditure. If consumer spending rises faster than the economy's ability to increase the supply of goods and services, the result can be upward pressure on prices.
This effect may be particularly relevant in the US, where household ownership of equities is relatively high and AI-related gains have contributed to a sharp increase in stock market wealth[19]. While the magnitude of the wealth effect is difficult to quantify, stronger demand generated by rising asset prices represents another channel through which the AI boom could add to inflationary pressures in the near term.
Inflation today, disinflation tomorrow?
All of this suggests that, for now, AI is exerting upward rather than downward pressure on US inflation. While AI-driven productivity gains hold out the prospect of a positive supply shock that could eventually lower costs and ease price pressures, there is little sign of such effects in the current data. Major technological innovations often take years to diffuse through the economy and translate into measurable productivity improvements.
By contrast, the demand-side effects of the AI boom are already visible. Massive investment in data centres is driving up the prices of computer chips, energy and raw materials, while rising equity prices may also support consumer spending through positive wealth effects. Both factors have the potential to put further upward pressure on inflation.
Against this backdrop, Chair Warsh will have a hard time convincing his Federal Open Market Committee (FOMC) colleagues that AI is going to bring down inflation anytime soon, and therefore provide the justification for near-term interest rate cuts. If anything, the AI boom currently appears to strengthen the case for keeping monetary policy tighter than it otherwise would be.
19 June 2026
[3] https://www.bls.gov/news.release/pdf/prod2.pdf
[4] https://www.cfodive.com/news/ai-cited-top-reason-us-job-cuts-third-straight-month/822029/
[5] https://www.dallasfed.org/research/economics/2026/0224
[6] https://tradingeconomics.com/united-states/non-farm-payrolls
[7] https://fred.stlouisfed.org/graph/?graph_id=1603406
[8] https://tradingeconomics.com/united-states/inflation-cpi
[9] https://tradingeconomics.com/united-states/core-inflation-rate
[10] https://www.afhwm.co.uk/news/how-could-the-fed-change-under-kevin-warsh-and-what-might-it-mean/
[11] https://fred.stlouisfed.org/series/TRMMEANCPIM159SFRBCLE
[12] https://uk.finance.yahoo.com/news/ai-could-capex-1tn-cy27-102810343.html
[15] https://fred.stlouisfed.org/series/CUSR0000SEEE01#
[16] https://fortune.com/2026/05/04/is-ai-inflationary-goldman-sachs-gen-z/