Capital, electricity, and talent can’t keep up! U.S. manufacturing is losing to AI.

The revival of U.S. manufacturing is facing an unexpected competitor—the AI data center boom is squeezing traditional manufacturing across the three key areas of capital, electricity, and labor, threatening the core policy goal of the Trump administration to rejuvenate American industry.

On October 24, according to Bloomberg, while the Trump administration is flying the banner of "manufacturing reshoring," a fiercer battle for resources is playing out behind the scenes. Tech giants are expected to invest up to $4 trillion in AI infrastructure by 2030. This investment frenzy, comparable to the railroad construction boom of the 19th century, is siphoning capital, electricity, and labor away from traditional factories to AI data centers at a frenzied pace.

Since the start of this year, data center construction spending has surged 18%, while new factory construction has shrunk by 2.5%. In regions dense with data centers, wholesale electricity prices have skyrocketed 267% over five years. Of the national shortage of 439,000 construction workers, one-fifth of contractors are now busy with data center projects. Even more ironically, tech companies are easily granted tariff exemptions, while manufacturers face the heaviest tax burdens since the 1990s for the equipment needed to expand factories.

The story of Lordstown, Ohio, epitomizes the absurdity of this transformation. While Trump promised to "bring jobs back" here in 2017, the General Motors plant was still shuttered. Now, SoftBank has taken over the facility, not to make cars, but to produce data center equipment—this former car manufacturing base, which once employed 12,000 workers, is now being converted by Foxconn, OpenAI, and SoftBank into a center for AI equipment manufacturing and demonstration data centers, expected to employ only about 1,600 people.

Research institute Pantheon Macroeconomics estimates that without AI-related infrastructure spending and the wealth effect from soaring AI stocks, U.S. GDP growth for the first half of 2025 would be only 1% instead of the actual 1.6%. Bloomberg Economics projects that, as Google, Meta, and Microsoft raise AI capital expenditure from nearly $400 billion this year to $600 billion next year, AI could contribute as much as 1.5 percentage points to GDP growth next year.

MIT researcher Paul Kedrosky warns that while manufacturing is struggling with tariffs, electricity prices, and labor shortages, AI enjoys policy green lights, capital fervor, and preferential resource allocation. The enthusiasm for AI technology has "broken people's mental model of how the economy works," causing policymakers to overlook the negative impacts of their policies.

Imbalanced Capital Flows: Data Center Investment Surges, Factory Construction Shrinks

The AI investment frenzy is causing a severe skew in capital allocation. According to Bloomberg Economics, the main players are expected to invest up to $4 trillion in AI infrastructure by 2030, on par with the railroad boom of the 1870s and the late 1990s fiber network frenzy.

This concentration of funds is evident in the data. Year to date, spending on data center construction is up nearly 18%, while spending on new factory construction is down 2.5%. According to the Institute for Supply Management, factory activity has contracted for seven straight months through September.

Morten Wierod, CEO of Swiss manufacturer ABB, points out that with tariffs raising costs and a shortage of construction workers, returns for builders and suppliers on data center projects are currently much higher than for other projects.

In October, ABB agreed to sell its industrial robotics business to SoftBank Group for over $5 billion to deepen its focus on the more lucrative data center business.

John Engel, CEO of electrical equipment distributor Wesco International, remarked:

"AI is consuming most of the oxygen in the room, and that's where the growth is right now. If you're not involved in it somehow, unfortunately, you're missing out."

Power Struggle: Data Center Demand Drives Up Industrial Electricity Costs

Electricity supply is quickly becoming the real limiting factor in whether the U.S. can simultaneously support the AI revolution and manufacturing revival.

According to the International Energy Agency, a typical AI data center consumes as much electricity as 100,000 households, and the largest data centers under construction will consume 20 times that. Bloomberg Industry Research estimates that by 2032, data centers could account for 20% of U.S. electricity demand.

Bloomberg News analysis has found that in areas near data centers, wholesale electricity costs have risen by as much as 267% over the past five years. This surge in demand is pushing up utility bills for both residential and industrial customers.

The Trump administration’s determination to cancel all federal support for renewables has further strained electricity supply.

To meet the huge energy needs of AI technology, utilities are investing in gas turbines. The data center construction boom has also driven up demand for all kinds of electrical equipment—Eaton Corp. reports second-quarter orders from data center customers were up 55% from last year.

Worsening Labor Shortage: Construction Workers Siphoned by AI Projects

The U.S. has already faced a long-term shortage of skilled workers, and Trump’s crackdown on illegal immigration has further exacerbated the gap. According to the Associated Builders and Contractors (ABC), the U.S. will be short 439,000 construction workers this year.

AI infrastructure construction is worsening the crunch. Of ABC’s 23,000 member companies, one-fifth signed data center project contracts in September.

John Fish, CEO of Suffolk Construction, says his company is building more than 30 data centers in eight states and it’s “tough” to find enough plumbers, HVAC contractors, mechanics, and electricians nationwide. Fish warns:

"If we don't take action to address this, the U.S. workforce will face a catastrophic problem. The economy is only going to need more and more of these skilled workers, and we are headed in the opposite direction."

Policy Contradictions: Selective Tariff Exemptions

The Trump administration’s policies toward AI and traditional manufacturing contrast sharply. Tech giants have successfully lobbied for broad tariff exemptions on imported servers and other data center hardware, but the government has been lukewarm in response to manufacturers’ requests for tariff exemptions on imported equipment needed to expand or build new U.S. factories.

In September, the Commerce Department began investigating imports of robots and industrial machinery, which could lead to more tariffs, further hindering reshoring efforts.

It’s estimated that Trump’s import tariffs amount to the largest tax hike on U.S. businesses since the early 1990s. Tariffs will cost Caterpillar up to $1.8 billion this year, while GM faces a tab of $4.5 billion.

Barclays analyst Julian Mitchell wrote in an August report that the industrial recovery "looks suspect" and called AI the "only game in town."

The National Association of Manufacturers has publicly supported the "Big, Beautiful Bill" passed by Congress this summer, which includes improved tax treatment for equipment purchasing and R&D investment.

Nonetheless, a survey of its members found that nearly 80% of companies named tariff burdens as their top concern. Based on these responses, the association predicts capital investment will grow only 1% (not inflation-adjusted) over the next year, in line with recent years.

Lordstown as a Symbol: From Car Manufacturing to AI Manufacturing

The transformation of the former GM plant in Lordstown has become a microcosm of this economic shift.

In July 2017, Trump told residents at a rally in nearby Youngstown, “Don’t sell your house” because “we’re bringing these jobs back, filling up those factories, or tearing them down and building brand new ones." GM closed the plant less than two years later, and a startup moved in to make electric cars, but went bankrupt by 2023.

In August, Foxconn announced it would sell the 6.2 million square foot facility for $375 million to a mysterious buyer, later revealed as SoftBank. The two companies plan to manufacture data center equipment at the location that once made Chevy Impalas and other models. SoftBank will also collaborate with OpenAI to test “advanced data center designs” at the former auto plant site.

The project will commence operations in 2026 and employ about 1,600 people, including the 400 already working there. For a region that has lost 40% of its manufacturing jobs in the last 20 years, these jobs matter, but they fall far short of filling the gap.

Some local leaders are cautious. Foxconn once promised, during Trump’s first term, to build a $10 billion LCD factory complex in Wisconsin—celebrated by Trump as the “Eighth Wonder of the World”—but it never materialized. Former Lordstown Mayor Arno Hill commented:

"They say this is the wave of the future, but will the market become saturated? People will always need cars."

MIT Digital Economy Initiative researcher Paul Kedrosky was among the first academics to put the AI capital expenditure boom in a historical context.

He worries that the technology fervor has blinded Trump and his advisers to the negative impact of their policies on other parts of the economy. "This thing is so big that it’s broken people’s mental models of how the economy works, and as a result, they are making mistakes," he said.

Risk warnings and liability disclaimerThe market has risks, so investment should be cautious. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation or needs of any individual user. Users should consider whether any opinions, views or conclusions in this article apply to their particular circumstances. Investing accordingly is at your own risk.