The data center buildout required to power artificial intelligence is generating inflation pressure today, while the cost reductions tech leaders promise depend on corporate adoption that has been slow to materialize. The expensive side of the AI trade is in current price data. The savings side is not.
The deflationary argument tech leaders are making
Tech leaders have argued that artificial intelligence will drive costs lower. Businesses deploying AI tools at scale, the reasoning goes, gain enough productivity to eventually compress prices across the economy. If adoption reaches the pace the optimist case requires, the Federal Reserve would receive disinflationary pressure from the private sector at exactly the moment it needs it. The case has internal logic. It also carries a timing condition: adoption must happen, and happen broadly, before any cost savings show up in measured prices. Corporate AI adoption has been slow.
Why construction costs land in the price level now
Data center construction is capital-intensive. Those costs flow into the price level immediately, before most enterprises have actually deployed the AI software those data centers are built to support. The inflation from the build is present-tense. The savings are downstream of adoption. Adoption has not arrived at scale.
What slow adoption means for the Fed
The Federal Reserve targets measured inflation, not projected deflation. With corporate adoption lagging, the mechanism that would eventually offset data center costs is broader AI productivity gains across businesses. That mechanism has not yet produced anything visible in current data. Meanwhile, capital expenditure on data centers continues. Tech leaders may be right about where costs eventually go. The data center buildout, right now, is moving in the opposite direction.