A data center can make a business more productive while making electricity more expensive for everyone around it. That is the complication behind the Fed’s growing attention to AI investment.
In her September 28 speech, Federal Reserve Governor Lisa Cook distinguished between rising prices within AI-related industries and inflation spreading across the economy. She cautioned that using monetary policy to fight a narrow sector’s price increases could be a mistake. Interest rates are too blunt for that job.
The policy problem changes when the inputs are shared.
Data centers compete for construction workers and energy that other businesses also need. Cook warned that this could transmit costs beyond technology companies, even as inflation within the AI sector itself moderates. Our earlier examination of AI’s electricity requirements helps explain where that transmission begins.
Dallas Fed research gives the electricity channel a measurable scale. Under its hypothetical scenarios, the data center boom could add approximately 0.04–0.13 percentage points to annual PCE inflation in 2030. Those are conditional estimates for electricity’s contribution, rather than a forecast of AI’s total inflation impact.
Our reading is that falling chip prices alone would offer an incomplete reassurance. AI hardware could become cheaper while the cost of powering and constructing the facilities continues feeding into other industries.
For gold, that creates a specific risk: broader inflation pressure could warrant tighter monetary policy. If that translates into higher real yields, holding a metal that pays no interest becomes more costly relative to inflation-protected bonds.
The boundary to watch is how far the costs travel. A price increase confined to AI equipment presents a different policy challenge from one reaching businesses and households that never ordered a server.
Leave a Reply