AI Data Center Demand Causes U.S. Manufacturing Energy Costs to Skyrocket

Manufacturers in the Rust Belt region of the U.S. Midwest are grappling with skyrocketing energy costs caused by the rapid proliferation of AI data centers. This situation could have a significant impact on former President Donald Trump’s “Made in America” initiative aimed at revitalizing domestic manufacturing. According to a report by Ars Technica, electricity prices for factories in the 13 states overseen by power grid operator PJM Interconnection have been rising at a considerably faster pace than those for other commercial and residential users.

Background

The Trump administration made the repatriation of manufacturing one of its top policy priorities, aiming to boost domestic production in sectors like steel, cement, and automobiles through measures such as tariff increases and deregulation. At the same time, the administration has actively promoted the construction of AI data centers. To compete with China in the AI race, the government has supported the development of large-scale computational infrastructure.

However, these two policies are beginning to clash over the shared resource of the power grid. Data centers consume vast amounts of electricity, ranging from several megawatts to tens of megawatts per facility. States in the Rust Belt have been attracting data centers with tax incentives and land offers, but ironically, this success is now putting pressure on traditional manufacturing industries.

PJM Interconnection operates the largest wholesale electricity market in the U.S. The construction boom of data centers in the region has led to a surge in predicted electricity demand, driving up capacity prices across the grid. According to a Reuters analysis, the price of capacity per megawatt-day is expected to rise from $28.92 in 2024 to $329.17 in 2026—a more than tenfold increase.

The Reality of Rising Energy Costs

One example is the Belden Brick Company, a longstanding brick manufacturer in Ohio with a 141-year history. The company’s monthly electricity costs have ballooned from $1,600 to $12,000—an increase of approximately 7.5 times. The primary reason is the hike in capacity charges, which are designed to distribute the costs of maintaining power plants among consumers. The surge in data center demand has strained supply reserves, directly driving these charges higher.

The impact on the steel industry is even more severe. The Steel Manufacturers Association has warned that steelmakers concentrated in the PJM region are bearing additional energy costs amounting to tens of millions of dollars annually. Energy expenses account for 20–40% of steel production costs. The power demand of individual electric arc furnaces ranges from 40 to 200 MW, and the U.S. steel industry as a whole consumes up to 11 GW of electricity during peak times.

Metallus, a steel manufacturer in Ohio, reported a 70% increase in energy costs since 2024, with an annual additional burden of $15 million. According to an Ars Technica article, while the construction of data centers generates demand for approximately 1 million tons of steel annually, it simultaneously raises operational costs, creating a double-edged sword for the industry.

PJM predicts that by 2027, electricity demand will exceed supply by 6.6 GW in its jurisdiction—a shortfall greater than the capacity of six nuclear power plants. The Wall Street Journal also reported that strained power grids have prompted steel industry executives to warn of potential blackouts.

Impact on Manufacturing

In response to rising energy costs, some manufacturers are passing the added costs onto product prices. However, industries exposed to international competition struggle to do so. Reuters reports that some companies are considering relocating their factories to regions with lower energy costs, such as the southern U.S. or overseas. Such relocations could lead to job losses and economic stagnation in the Rust Belt.

The steel industry is particularly concerned that grid strain increases the risk of production outages. During peak summer periods, competition for limited electricity between data centers and factories becomes especially acute. Because data centers require 24/7 operational stability, grid operators may prioritize their needs over manufacturing facilities, potentially cutting off power to factories.

This situation also creates a conflict with the Trump administration’s tariff policies. While tariffs aim to limit imports of steel, rising domestic production costs could erode price competitiveness, ultimately impacting downstream industries like automotive and construction materials.

The Trump Administration’s Dilemma

During his election campaign, former President Trump pledged to bring manufacturing jobs back to the U.S. under the banner of “Made in America.” Simultaneously, the administration has promoted data center construction to maintain the country’s status as a global AI leader. Reporting from Ars Technica highlights how these two policies are beginning to interfere with one another.

Within the administration, discussions are reportedly underway to accelerate the approval of power plant construction by the Department of Energy and to review PJM’s market design. However, building new power transmission lines can take decades. Existing fossil fuel power plants are being phased out due to environmental regulations, and the construction of new nuclear plants is facing delays. Short-term solutions are limited.

The data center industry is also under pressure to adapt. Major hyperscale companies have announced investments in renewable energy, but the lead time for connecting such resources to the grid remains long. Immediate expansion of natural gas-fired power plants would likely face strong opposition from environmental groups.

Future Outlook

The electricity market in the PJM region is expected to face a chronic supply shortage starting in 2027. High capacity prices are likely to persist, potentially altering the cost structure of manufacturing industries permanently. Projections from the steel industry suggest that the viability of existing factories could be at risk.

In the medium to long term, expanding power transmission lines and introducing distributed energy resources will be critical. Data centers may accelerate investments in battery storage and on-site power generation. However, these costs will ultimately be passed on to the prices of AI services.

Analysis from Ars Technica indicates that, much like in semiconductor manufacturing, the U.S. will need to optimize resource allocation between AI infrastructure and manufacturing industries. The competition for electricity—a critical resource—presents an opportunity to reconsider policy priorities from the ground up.

Editorial Opinion

In the short term, rising electricity costs in the Rust Belt seem inevitable as the data center construction boom continues. If the anticipated power supply shortages become a reality by 2027, the declining competitiveness of the manufacturing sector will become more apparent. Investors may need to revise their earnings projections for industries sensitive to energy costs.

From a long-term perspective, achieving coexistence between the AI industry and manufacturing will require significant expansion of the power grid. However, given the challenges of securing permits and land, this could take a decade or more. Without the adoption of renewable energy or small modular nuclear reactors, the trade-offs between these two sectors will only deepen.

The editorial team believes that balancing the promotion of AI and the protection of manufacturing is not merely a policy slogan but a practical challenge that must contend with the physical limitations of the power grid. The Trump administration must clearly prioritize between incentivizing data center development and supporting manufacturing. This issue also holds implications for Japan’s policies on attracting data centers.

References

• Data centers’ energy demand threatens Trump’s “Made in America” plan - Ars Technica — Published on 2026-07-07