Does AI exacerbate U.S. inflation? Goldman Sachs: Memory, electricity and software price increases may raise core PCE by 0.5 percentage points by the end of the year.

The artificial intelligence boom is reshaping the inflation landscape in the United States in unexpected ways. Goldman Sachs economist Megan Peters calculated in a recent report that the threefold effect of AI-driven memory price spikes, software price increases, and electricity price rises has already raised the core PCE year-on-year inflation rate in the United States by more than 0.2 percentage points, with this contribution expected to rise to 0.5 percentage points by the end of the year. This estimate does not fully reflect various spillover effects, and the actual impact may be larger. This warning has resonated within the Federal Reserve. The latest minutes from the Federal Open Market Committee (FOMC) meeting explicitly raised concerns that "strong AI-related demand is leading to persistently high inflation." The Federal Reserve's semi-annual monetary policy report also lists "increased demand for high-tech products supporting AI applications" as one of the factors pushing prices higher. Internal divisions within the Federal Reserve The issue of AI inflation has sparked rare public disagreements within the upper echelons of the Federal Reserve. New York Fed President John Williams stated in a speech that if AI demand "continues to have a sustained impact on supply and demand, pushing up inflation, I believe this situation should not be ignored." This stance is in stark contrast to that of Federal Reserve Chairman Jerome Powell. In November of last year, Powell wrote an article characterizing AI as an "important deflationary force," believing that it would enhance U.S. competitiveness through increased productivity. Although this assessment may have some basis in the long term, it is undeniable that the skyrocketing prices of memory have already had a cost impact on consumer electronics products. Three major channels driving core inflation upward Goldman Sachs quantified the transmission pathways of AI on inflation from three dimensions. Memory prices are the most direct channel of impact. Since the beginning of 2025, the strong demand for data centers has driven some memory prices up more than tenfold. Goldman Sachs predicts that the software and accessories price index highly correlated with memory prices will peak at a year-on-year increase of about 30% in November, contributing around 36 basis points to core PCE year-on-year. Recent price increases of up to 25% announced by major consumer electronics manufacturers such as Apple, Microsoft, and Dell are a direct reflection of this trend. While memory price increases are a global phenomenon, the degree of impact varies by country. In non-U.S. developed markets such as the Eurozone, the United Kingdom, and Canada, the peak contribution of inflation categories directly affected by memory is about 1 to 9 basis points, averaging about 5 basis points, significantly lower than the United States - primarily due to the higher weight of related items in the U.S. PCE basket. The second channel is software price increases. In January 2025, Microsoft raised the price of consumer Microsoft 365 subscriptions for the first time since introducing the subscription model in 2013, citing the introduction of the AI Copilot feature. UK statistics show that in March 2025, software prices saw a quarterly increase of 20%, the highest on record. Since the weight of software in the inflation basket of most developed markets is very low, the impact outside the United States is relatively limited - in the UK, a 20% overall software price increase contributes less than 4 basis points to core inflation, even lower in other regions. The third transmission channel is the rise in electricity prices. The increasing electricity demand from data centers has started to push up residential electricity costs in the United States, with the impact particularly prominent in areas with dense data center concentration. Goldman Sachs estimates that the current electricity demand from data centers has contributed about 8 basis points to core PCE year-on-year inflation through the electricity channel, with some PJM grid states showing higher impacts than the national average. Continued rise in data center electricity demand The medium to long-term trend of electricity prices is also not optimistic. According to Goldman Sachs' research, the share of electricity used by data centers in the total national electricity consumption in the United States is expected to increase from about 6% currently to 11% by 2030, nearly doubling. In contrast, the share of electricity used by data centers in the EU has been stable at 3-4% in the long term, and is expected to increase only slightly to about 6% by 2030, indicating significantly lower pressure. A report cited by Goldman Sachs shows that in Ireland, where data centers account for as much as 23% of electricity usage, the construction of data centers has cumulatively raised the energy bills of ordinary households by around 360 between 2015 and 2023. This case may provide a reference for the future trend in the United States. For medium-sized developed markets and emerging markets with densely concentrated data center construction layouts, there is still significant room for further pressure on electricity prices. Policy dilemma: The Federal Reserve finds it difficult to "reverse" memory prices The above analysis puts the Federal Reserve in a tricky situation. When core PCE inflation faces additional upward pressure from supply-side technological changes, monetary policy tools have no direct influence on memory prices, electricity prices, or software subscription costs. At the same time, there is no sign of a slowdown in the current AI capital expenditure boom, which means that related inflationary pressures will continue in the foreseeable future. For market participants, if Goldman Sachs' calculations come true and core PCE inflation contribution reaches 0.5 percentage points by the end of the year, it will further narrow the Fed's window for interest rate cuts this year, and may force policymakers to re-examine their reliance on the narrative of long-term deflation due to AI. This article is from "Wall Street News," Author: Yang Chen, GMTEight Editor: Li Cheng