NEW YORK — Investors and strategists say AI-driven inflation is becoming a sleeper risk for 2026 as companies pour money into power-hungry data centers and scarce, high-end chips. The concern is that these inputs are turning into bottlenecks that keep price pressures hotter than markets expect, Jan. 6, 2026.
In a Reuters report, Morgan Stanley strategist Andrew Sheets put it bluntly: “The costs are going up not down in our forecast, because there’s inflation in chip costs and inflation in power costs.” Others warn that if inflation re-accelerates, central banks could end rate cuts sooner than investors are betting, tightening financial conditions for the very AI buildout driving the boom.
The mechanism is simple: hyperscalers can absorb higher bills, but their supply chains cannot expand overnight. When construction crews, transformers, turbines, and advanced memory are all in short supply at once, AI-driven inflation can show up first in corporate budgets — and then in the prices households pay for everything from cloud services to consumer electronics.
AI-driven inflation: the new bottleneck economy
Power is the first pinch point. The U.S. Energy Information Administration projected electricity consumption would hit record highs in 2025 and 2026 as data centers ramp up, according to a June 2025 Reuters dispatch. In practice, that translates into utilities racing to add capacity, upgrade transmission, and accelerate interconnection work — efforts that can lift costs long before new electrons arrive.
Globally, the growth curve is steep. The International Energy Agency said electricity demand from data centers is set to more than double by 2030, driven largely by AI workloads, in an April 2025 update. In the U.S., S&P Global’s 451 Research forecast that grid power demand from data centers would jump sharply in 2025 and keep climbing through the decade, as detailed in a 2025 S&P Global analysis.
The second pinch point is chips — especially the memory and accelerators needed to train and run large models.
A parallel shortage is already playing out in real time: another Reuters report this week described a supply crunch tied to demand for AI infrastructure, with some memory prices more than doubling from prior-year levels. That feeds directly into server costs, then into cloud pricing and enterprise IT budgets — another channel for AI-driven inflation.
This is not the first time chips have driven broader price swings.
During the pandemic-era crunch, the Cleveland Fed documented how semiconductor shortages constrained vehicle output and coincided with rising new-car prices in a 2021 economic commentary. The St. Louis Fed also traced how scarcity and production bottlenecks can filter into inflation readings in a 2022 post.
What’s different in 2026 is the scale and simultaneity:
power, construction, and high-end semiconductors are tightening together, amplifying AI-driven inflation risk. Investors say the tell will be whether productivity gains from AI arrive fast enough to offset the near-term surge in infrastructure spending — or whether the buildout simply pushes costs through the system before efficiencies kick in.
