US Data Centers Encounter Multiple Challenges Amid Grassroots Opposition to AI

The recent surge in demand for datacenters in the United States, driven primarily by the artificial intelligence (AI) revolution, has been met with a series of disruptions that have slowed down or even halted new developments. These setbacks stem from various challenges, including supply chain disruptions, energy shortages, and regulatory hurdles linked to tariffs. As the landscape evolves, stakeholders must navigate a complex array of barriers to harness the potential of AI’s growth.

Grassroots opposition has emerged as a significant factor impacting the construction of new facilities, with local communities voicing concerns that have led to project cancellations. Additionally, investors are displaying increasing apprehension regarding datacenters, partly due to fears of an impending AI bubble.

Reports from investment research firms, notably MacroEdge and climate-focused outlet Heatmap, indicate that numerous datacenter plans have faced significant delays or outright cancellations throughout December and January. MacroEdge, for example, recorded 26 cancellations by the end of January—an alarming rise compared to just one cancellation in October.

This compilation of challenges calls into question the United States’ ability to effectively manage and expedite the datacenter expansion that has been critical in fostering economic growth over the last 18 months. Delays in production and infrastructure could potentially have significant repercussions on the broader economy, as highlighted by MacroEdge’s chief economist, Don Johnson.

Johnson noted, “The [Trump] administration is going to be scrambling to find its next growth engine as the datacenter machine winds down as a tailwind.” This statement underscores the interconnected nature of datacenter development and national economic health.

Across the U.S., numerous proposals for hyper-scale datacenters—designed to host infrastructure supporting AI—are on the table. These facilities have massive energy demands, rivaling those of major cities. Consequently, energy grids must evolve rapidly, which entails integrating new components such as transformers, circuit breakers, high-voltage cables, and robust steel poles to connect these datacenters to the existing power supply.

Marsden Hanna, Google’s head of energy and sustainability, emphasized the hurdle of grid connectivity during a recent industry conference, stating, “Connecting to the grid is the No 1 challenge we’re seeing.” He elaborated that utility companies in numerous markets estimate connection timelines ranging from four to ten years, with some indicating that the interconnection study alone could take up to 12 years.

The challenge of tracking both proposals and cancellations is complicated by the absence of a central oversight agency. Douglas Jester, managing partner at 5 Lakes Energy—a firm focused on regulatory matters around datacenter construction—pointed out the convoluted nature of the planning process. Datacenter developers, including major players like Google and Oracle, must engage multiple regional utilities to assess costs and timelines for grid connection.

Since developers frequently submit identical plans in various locations across the country, it becomes arduous to ascertain when a project is genuinely active. Nevertheless, the accumulation of obstacles related to energy supply remains a pressing concern. Many energy grids lack the capacity to meet immediate demands, and the sluggish nature of augmenting power supply adds another layer of delay.

Regional grid operators, responsible for coordinating energy generation and distribution, often require up to five years to evaluate the impact of new gas plants, solar fields, or other power generation methods on the grid. This protracted review process is already impeding the transition to clean energy—exemplified by the fact that achieving equivalent power output from solar installations necessitates significantly more resources compared to traditional gas plants.

Jester noted, “The interconnection process is really getting bogged down, and it has been a problem even before datacenters.” The reviews ensure that new energy sources won’t disrupt the grid supply; however, Jester highlighted that the Electric Reliability Council of Texas (Ercot) adopts a more efficient approach, swiftly incorporating new energy generation before addressing any issues that arise.

The recent wave of datacenter projects has exacerbated existing challenges related to energy grid supply chains, which are still recovering from disruptions caused by the COVID-19 pandemic and facing strain from the clean energy transition, according to Qiuhua Huang, an electrical engineering associate professor at Colorado School of Mines. Demand for essential equipment, such as transformers, has surged, yet the U.S. has only one production facility capable of supplying the required type of steel. Copper shortages further compound the situation.

High-voltage transformers, which previously had lead times of around six months, now take as long as four years to manufacture. The industry also grapples with a shortage of skilled labor, compounding delays and difficulties in scaling up operations.

Historically, the U.S. utility sector has depended heavily on imports for utility infrastructure, particularly from China and other countries. However, the global demand has surged, leading to inflated prices in the U.S. Furthermore, tariffs implemented during the Trump administration have made the U.S. market less appealing to foreign suppliers, resulting in transformer prices that have skyrocketed up to six times compared to pre-2022 levels.

Jester remarked, “The tariffs are exacerbating, if not causing, the problem in the U.S.” This increasing grid complexity extends the timelines for new datacenter projects, with PJM—the largest grid operator—pushing connection timelines back to as far as 2030, according to MacroEdge.

Johnson urged caution amongst investors in light of these challenges. Some hedge funds and datacenter investors are indeed becoming increasingly apprehensive. For example, Blue Owl, which had invested in the controversial Saline Township datacenter in Michigan, withdrew its $10 billion investment last December. Although Oracle insists it will progress with the project, similar hesitations have been noted from firms like Bain Capital and others.

Solutions are gradually emerging to alleviate some of these pressures. Steel manufacturers are ramping up production, and battery storage systems are becoming an increasingly viable alternative to new power generation. Jester suggests that integrating large-scale battery storage into the grid can facilitate energy retention for future use, thereby mitigating the need for new power plants and extensive grid updates.

Oracle has proposed that such battery storage could help it achieve its goal of developing a 1.4GW datacenter in Michigan by 2027, a project that would demand energy equivalent to that consumed by Detroit. DTE Energy, which serves the area, is nearing grid capacity and likely cannot install sufficient solar fields or gas plants to satisfy this demand within the projected timeline due to the sluggish review process.

In the backdrop of these challenges, major tech firms, armed with greater resources than traditional utility providers, are innovating to develop new transformer technologies that require fewer raw materials and less infrastructure. Huang noted, “It’s in the early phase of mass adoption, but datacenters have a much larger budget than utilities, so they can better handle those technology adoptions, and that will diversify the supply chain.”