Wells Fargo strategist Ohsung Kwon recently highlighted that technology giants' AI spending is "trickling down" into the broader real economy, with industrial stocks poised to be major beneficiaries. This assessment is not a vague optimistic forecast—through three dimensions of macro data, corporate earnings reports, and policy dynamics, the spillover effects of AI investment are genuinely materializing, yet they also face increasingly severe political headwinds.
The spillover effect is not just rhetoric: data is validating Kwon's judgment with solid foundations. According to Wells Fargo estimates, manufacturing activity expanded at its fastest pace in over four years in July, non-AI-related capital expenditure grew 10% year-over-year, and the pace of commercial and industrial loan growth has also accelerated notably. The S&P 500 industrial sector has risen 20% year-to-date, trailing only the energy and information technology sectors. A broader macro perspective confirms this trend. According to ConstructConnect data, spending on data center construction starts in the first five months of 2026 in the U.S. reached $58.1 billion, more than four times the same period in 2025; first-quarter spending on starts was $46.5 billion, up over 500% year-over-year. The AI boom has fundamentally reshaped the U.S. economy, with hyperscale cloud service providers expected to spend approximately $750 billion this year. IDC projects global AI infrastructure spending will reach $497 billion in 2026, growing about 56% year-over-year.
Who is benefiting? From Caterpillar to Vertiv, order backlogs are surging. Companies that "sell shovels" are reaping the rewards first. Caterpillar reported its quarterly earnings on Tuesday, with revenue exceeding $200 billion for the first time, up 24% year-over-year, driven by demand for power generation equipment and construction machinery fueled by data center builds. Sales in its construction industries segment surged 35%. The company subsequently raised its full-year revenue growth guidance. Data center infrastructure suppliers are also benefiting. Vertiv Holdings Co's backlog has surpassed $15 billion, with fourth-quarter orders surging 252% year-over-year; Eaton Corporation plc's electrical segment saw data center revenue rise approximately 50% year-over-year in the first quarter. Wells Fargo itself estimates that hyperscale cloud service provider capital expenditure will reach $1.1 trillion by 2027, about 25% above market consensus. Kwon estimates that roughly 40 mega-data centers are currently under construction in the U.S., with over 100 in planning stages, concentrated in Texas, Georgia, Virginia, and Pennsylvania.
Hidden concerns: Political resistance is becoming the biggest variable. However, Kwon explicitly warns: "The biggest risk to data center construction is political backlash, especially as the midterm elections approach." This risk is rapidly becoming more apparent. A Gallup survey shows 71% of Americans oppose building AI data centers in their local areas; 77% worry AI will drive up electricity prices. In July, opponents staged 142 protests across 42 states in the U.S. In the first quarter alone, local groups blocked or delayed 75 projects, involving investments of approximately $130 billion. Policy-level pressures are even greater. New York State signed the nation's first one-year moratorium on large AI data centers in July; at least 12 state gubernatorial candidates in the midterm elections have expressed support for pausing data center construction. Tax incentives are also rapidly fading—four states have canceled or suspended data center tax breaks, with nine others studying the issue. At a 7% sales tax rate, the equipment procurement cost for a single 1GW data center would increase by about $3 billion. Although the Trump administration introduced a "Electricity Ratepayer Protection Commitment," inviting tech giants to sign an agreement not to pass grid upgrade costs onto residents, the commitment is non-binding, and whether it can ease public backlash remains uncertain. Wells Fargo's assessment reveals an ongoing reality: the dividends of AI spending are spreading beyond semiconductor and cloud service providers, injecting growth momentum into traditional industrial sectors. But the sustainability of this narrative increasingly depends not on technology or capital, but on politics—how the November midterm elections reshape the regulatory landscape for data centers will be the key variable determining how far the "trickle-down effect" can go. As Kwon stated, we may still be in "very, very early stages," but early stages are often when the most volatility occurs.
Comments