AI Infrastructure Spending Creates Inflation Headwinds for Fed

The artificial intelligence industry's leaders have long promised that AI would drive down costs across the economy. OpenAI CEO Sam Altman recently wrote that "intelligence too cheap to meter is well within grasp," while Elon Musk has argued AI and robotics will create extreme abundance. SoftBank's Masayoshi Son predicted a 40% drop in prices.

Those predictions haven't materialized. Instead, the technology sector's multi-trillion-dollar spending spree on AI infrastructure is creating near-term inflation pressures while productivity improvements remain difficult to measure.

Why it matters

Federal Reserve officials must decide whether to raise interest rates based on current inflation data or hold steady in anticipation of future AI-driven productivity gains. The tension between immediate costs and promised benefits is dividing policymakers at a critical moment for monetary policy, with some Fed officials already dissenting in favor of higher rates to counter AI-related price increases.

The cost of building AI

Capital expenditure on AI infrastructure is expected to reach $581 billion this year in the United States alone—equivalent to 1.8% of GDP, according to Goldman Sachs Research. That figure could climb to 2.8% of GDP by 2028, with global spending approaching $1 trillion.

This massive buildout is creating tangible price pressures. Household electricity prices rose 10.1% in the two years through June, faster than the overall 6.3% inflation rate during that period. The cost of dynamic random access memory (DRAM) will have increased 400% by year-end compared to 2024, JPMorgan Chase estimates. Computer software and accessories have risen 22.9% since June 2024.

Supply chains for AI servers and chips are strained as companies compete for limited production capacity from manufacturers like Nvidia.

Adoption lags behind hype

A Census Bureau survey published in May found that only 17% to 20% of U.S. businesses reported using AI, with adoption concentrated among large firms. The gap between AI leaders and typical companies is widening rapidly.

Ronnie Chatterji, chief economist for OpenAI, said power users deploy AI at eight times the rate of average companies—up from a two-times gap three months earlier. "For it to impact the economy, it has to be adopted by organizations," Chatterji said. "Those organizations have to realize value."

Julie Averill, Lululemon's former chief information officer who oversaw AI adoption at the company, said implementation challenges persist. "The things that have always made implementations in large companies difficult still exist, which is people," Averill said. "Getting people to change their behaviors, taking them along the journey with you, and getting them to trust the model, that's hard."

The weak links problem

Economists studying AI use the term "weak links" to describe tasks that resist automation. Stanford professor Charles Jones, now on leave at Anthropic, points to radiologists as an example. Despite predictions in 2016 that AI would eliminate the profession within a decade, radiologist numbers kept growing as AI made them more productive at certain tasks while leaving others—like patient consultations—untouched.

Fed officials divided

Fed Chairman Kevin Warsh has assembled a task force including Jones and venture capitalist Marc Andreessen to advise on AI's economic effects. Warsh wrote in November that "AI will be a significant disinflationary force, increasing productivity and bolstering American competitiveness."

But other Fed officials remain skeptical. Minneapolis Fed President Neel Kashkari dissented in favor of higher interest rates in July, citing concerns that "the massive investment in data centers has also added a new demand element to the high inflation Americans are experiencing."

Warsh himself has adopted a more cautious stance, acknowledging in July that "the precise timing and magnitude of effects on the supply side remain hard to predict."

Peter Boockvar of One Point BFG Wealth Partners noted that even during the internet boom, U.S. productivity gained only 1.5% over 30 years. "To think that generative AI is going to bring that level of enhancement to the economy, relative to the internet, is tough," Boockvar said.

These details were first reported by CNBC.