AI Infrastructure Boom: The Hidden Driver of a New Round of Inflation?

With the Strait of Hormuz reopened, global oil prices have fallen sharply, briefly relieving the global economy from high inflation. However, as the market breathes a sigh of relief over the drop in energy prices, a more subtle and persistent new inflation variable is quietly emerging—the "investment tsunami" in artificial intelligence (AI) infrastructure. This capital flood, led by tech giants, is not only reshaping the global economic landscape but also posing unprecedented challenges to the Fed's already complex inflation management.

I. The "Siphon Effect" of Trillions in Capital

The explosive development of AI is far more than a technological breakthrough; it is an unprecedented capital-intensive industrial revolution. To seize the commanding heights of the future, global tech giants are racing to build AI infrastructure: from purchasing millions of dedicated AI computing chips to constructing high-energy data centers equipped with liquid cooling systems—every link consumes enormous sums of money.

According to TD Cowen's estimates, the capital expenditures of major hyperscale cloud service providers will reach a staggering $745 billion this year, and by 2027 and 2028, this figure will exceed the $1 trillion mark. Even more striking, the share of these top players' spending in GDP is expected to rapidly soar from less than 0.5% in 2020 to about 3% next year. This means that AI infrastructure is transitioning from a niche technology investment into a significant lever driving the macro economy.

This "investment tsunami" brings more than just on-paper numerical growth. It acts like a giant "siphon pump," drawing large amounts of capital, raw materials, labor, and even electricity from other industries and concentrating them into the AI supply chain. When the supply side cannot respond quickly in the short term, the explosive demand naturally translates into price pressure.

II. From Chips to Consumer Goods: The "Chain Reaction" of Inflation Transmission

The frenzy of AI infrastructure has already been clearly reflected in inflation data and is beginning to spread through the real economy to upstream and downstream sectors. The first to feel the pressure is the semiconductor and memory chip field. Advanced process capacity for AI chips is already tight, coupled with the insatiable demand from data centers for high-bandwidth memory, directly pushing up prices across the entire memory chip market.

The problem is that these memory and storage chips, "snatched away" by AI, are also core components for consumer electronics such as video games, cars, and smartphones. When supply is prioritized for AI, production costs for ordinary consumer goods rise accordingly. Apple announced this week that it will raise prices for iPads and MacBooks due to rising memory and storage costs—a vivid footnote to this transmission of inflation.

Moreover, AI infrastructure is also driving a sharp increase in demand for construction workers. From data center site selection to plant construction, vast projects require a large number of skilled workers. Against the backdrop of an already tight labor market, this additional demand further pushes up wage levels in the construction industry and may indirectly affect labor costs in other industries. Seemingly distant AI investments are quietly seeping into everyone's cost of living through hidden chains.

III. The "Two Sides" of the New Issue: Short-Term Inflation and Long-Term Deflation

Faced with this new inflationary pressure, divisions are emerging within the economics community. Oscar Munoz, TD Securities' Head of Economics, put it succinctly: "The initial phase of AI infrastructure build-out will be inflationary because demand puts pressure on a fixed economic supply side; but over the coming years, as productivity improves and the economic pie grows, it will eventually likely become a disinflationary force."

This view outlines a "two-phase" picture of AI's impact on inflation: the first phase is investment-driven inflation, with high capital concentration and resource scarcity; the second phase is productivity-driven deflation, where widespread adoption and integration of AI technology into existing processes will dramatically boost efficiency, lower unit costs, and ultimately benefit consumers through lower prices.

Greg Daco, Chief Economist at EY, echoed this view, noting that any technological revolution initially triggers inflation because the phase is "capital-intensive," with high demand chasing limited supply. From steam engines to the Internet, history has repeatedly proven that the pain of investment must be endured before reaping the benefits of technology. AI is no different—over the next one to two years, these inflationary pressures will continue to transmit to consumers until investment peaks and begins to decline.

IV. The Fed's "New Challenge": The Quiet Rise of the Neutral Rate

The emergence of short-term inflation is a further burden for the Fed, which is focused on controlling prices. More challenging is that the vast capital demand from AI infrastructure is quietly changing a key macroeconomic variable—the neutral interest rate.

The neutral rate is the benchmark interest rate at which the Fed neither stimulates nor restrains economic growth. Munoz clearly stated: "Even if AI infrastructure is only in the construction phase, it is pushing up the short-term neutral rate." This is because when trillions of dollars are concentrated in AI, the overall economy's demand for capital increases significantly, raising the cost of capital. To maintain economic balance, the Fed would need to keep interest rates at a higher level than in the past.

This change directly challenges the optimistic expectations previously expressed by Fed Chairman Kevin Warsh. Warsh had publicly said that AI would boost productivity, lower inflation, and give the Fed more room to cut rates. However, in the face of reality, this long-term vision may need reconsideration. In his most recent press conference, Warsh's language was already more restrained, emphasizing that "the Fed still has work to do on price stability" and not repeating the old refrain that AI would lower inflation.

V. Conclusion: The Era's Narrative Needs a Dual Dimension

Overall, the impact of AI infrastructure on inflation is far from simply "up" or "down." It is a complex, phased process. In the short term, resource tightness from massive investment will push up prices, adding greater uncertainty to the Fed's decisions; in the long term, if AI technology successfully proliferates and boosts total factor productivity, it could become a powerful deflationary force.

For investors, policymakers, and ordinary consumers, this means adjusting expectations: not to be complacent about falling oil prices or short-term inflation declines, nor overly pessimistic about price increases during the early phase of AI construction. True wisdom lies in understanding the dual narrative—acknowledging short-term pain while positioning for long-term gains.

The core challenge for the Fed is to find a balance between investment-driven inflationary pressure and potential future deflationary dividends. This requires not just precise economic modeling, but a deep understanding of the nature of technological revolution. The AI wave has arrived, and its impact is destined to be profound and complex. In this game, a mix of optimism and prudence is the most responsible answer to our times.