The artificial intelligence boom is reshaping the landscape of US inflation in unexpected ways.
Goldman Sachs economist Megan Peters estimates in a recent report that the combined effect of AI-driven memory price surges, software price hikes, and rising electricity costs has already added over 0.2 percentage points to the year-on-year core Personal Consumption Expenditures (PCE) inflation rate. This contribution is projected to increase to 0.5 percentage points by the end of the year. This estimate does not yet fully account for various spillover effects, meaning the actual impact could be even more significant.
This warning has found resonance within the Federal Reserve. The latest minutes from the Federal Open Market Committee (FOMC) meeting for the first time explicitly noted that most officials are concerned about a scenario where "inflation remains persistently elevated due to strong AI-related demand."
The Fed's semi-annual Monetary Policy Report also listed "increased demand for high-tech products that support AI applications" as one of the factors putting upward pressure on prices.
Divisions Emerge Within the Fed
The issue of AI-related inflation has sparked a rare public divergence among top Federal Reserve officials. In a speech, New York Fed President John Williams stated directly that if AI demand "creates persistent pressures on supply and demand and pushes up inflation, I don't think that's something to ignore."
This stance stands in clear opposition to the position of Fed Governor Michelle W. Bowman. In an article last November, Bowman characterized AI as a "significant disinflationary force," arguing it would enhance US competitiveness by boosting productivity.
While this assessment may hold some validity in the long term, the current reality is the undeniable pass-through of soaring memory prices to the cost of consumer electronics.
Three Channels Driving Core Inflation Higher
Goldman Sachs quantifies the transmission of AI to inflation through three primary channels.
Memory prices represent the most direct channel. Since early 2025, robust demand from data centers has driven prices for certain memory components up more than tenfold.
Goldman Sachs projects that the software and accessories price index, which is highly correlated with memory prices, will reach a peak year-on-year increase of approximately 30% in November. This is expected to contribute about 36 basis points to the year-on-year core PCE inflation rate. Recent price increases of up to 25% announced by major consumer electronics firms like Apple Inc., Microsoft Corporation, and Dell Technologies Inc. are a direct manifestation of this trend.
While memory price increases are a global phenomenon, their impact varies by country. In non-US developed markets like the Eurozone, the UK, and Canada, the peak contribution to inflation from categories directly affected by memory is estimated between 1 to 9 basis points, averaging about 5 basis points—significantly lower than in the US. This difference is primarily due to the much higher weight of these categories in the US PCE basket compared to other countries.
Software price hikes constitute the second channel. In January 2025, Microsoft raised the subscription price for its consumer Microsoft 365 plan for the first time since introducing the subscription model in 2013, citing the integration of its AI Copilot feature.
UK statistics show that software prices recorded their highest-ever quarterly increase of 20% in March 2025. Since software carries an extremely low weight in the inflation baskets of most developed markets, this effect is relatively limited outside the US. In the UK, a 20% overall software price increase contributes less than 4 basis points to core inflation, with an even smaller impact in other regions.
Rising electricity costs form the third transmission chain. Power demand from data centers has begun to push up residential electricity bills in the US, with the effect being particularly pronounced in regions dense with data centers.
Goldman Sachs estimates that data center electricity demand is currently contributing about 8 basis points to year-on-year core PCE inflation via electricity prices, with the impact in some PJM Interconnection states far exceeding the national average.
Surging Data Center Power Demand
The medium- to long-term outlook for electricity price pressure is also concerning.
Based on Goldman Sachs research projections, the share of US electricity consumption used by data centers is expected to rise from the current level of about 6% to 11% by 2030, representing a near-doubling of their consumption share. In contrast, the EU's data center electricity share has long been stable at 3-4% and is projected to rise only modestly to around 6% by 2030, indicating significantly lower pressure.
A report cited by Goldman Sachs shows that in Ireland, where data centers account for a substantial 23% of electricity consumption, data center construction cumulatively added approximately €360 to the average household energy bill between 2015 and 2023.
This case may provide a reference point for the future trajectory in the US. For smaller developed and emerging markets with concentrated data center construction, there remains significant potential for further pressure on electricity prices.
A Policy Conundrum: The Fed's Limited Ability to Counteract Memory Prices
This analysis places the Federal Reserve in a difficult position.
When core PCE inflation faces additional upward pressure from supply-side technological shifts, monetary policy tools have no direct influence over memory chip prices, electricity rates, or software subscription fees. Simultaneously, the AI capital expenditure boom shows no signs of abating in the near term, suggesting these inflationary pressures will persist for the foreseeable future.
For market participants, if Goldman Sachs's projections materialize and the year-end contribution to core PCE inflation reaches 0.5 percentage points, it would further narrow the window for Federal Reserve interest rate cuts this year. It could also compel policymakers to reassess their previous reliance on the narrative of AI as a long-term disinflationary force.
Comments