The Tech Energy Transition

Guest Blogger

For the Balkinization Symposium on the Global Political Economy of Artificial Intelligence.

Amy L. Stein 

It has been three and a half years since ChatGPT exploded on the scene, and many have yet to wrap their heads around the environmental, economic, and ethical implications of AI’s lifeblood – data centers. Without these sprawling facilities that house thousands of servers and mountains of computational equipment, AI is ineffectual. This makes the physical limitations of AI’s growth much more pronounced that initially anticipated. The true bottleneck for AI development is the energy required to power these data centers, rendering the electric grid ground zero for data center development disputes. These disputes center around the clash of two titans: Big Tech and Big Utilities. While the largest privately-owned electric public utilities (Big Utilities) have dictated the shape of our grid for the last century, Big Tech’s emergence in the grid space has disrupted assumptions, regulatory approaches, and ratepayer economics in ways previously unimaginable. This renders power dynamics in the electric grid space a ripe testing ground for political economy scholars focused on how power, wealth, and institutional rules determine winners and losers. 

My most recent work, Shifting Structural Power: The Tech Energy Transition, 101 Indiana Law Journal 583 (2026), draws upon theories of structural power in the international political economy (IPE) literature to tease out the relationships between these Big Utilities and Big Tech.[1] In the IPE literature, structural power is generally understood as the power to choose and to shape the structures of the global political economy within which other stakeholders must operate. Professor Susan Strange was concerned that financial wealth dictates politics, and that corporate influence had surpassed the control of governments. As she elucidated in 1987, power was not only about tangible resources like territory and military, but also about the power to decide how things would be done, to shape frameworks within which actors relate to each other, and to set the agenda in a way that defines the choices available to others.[2] She explained structural power as being found in four separate but interrelated structures: the ability to control (1) other people’s security; (2) the system of production of goods and services; (3) the structure of finance and credit through which it is possible to acquire purchasing power without needing to work or trade it; and (4) the acquisition, communication, and storage of knowledge and information. These four pillars are useful in identifying sources of power, organizing the mechanisms through which power can operate in policy processes, and exploring the consequences of such shifts for regulatory and political systems and outcomes. And they map nicely onto Big Tech’s activities in the grid space. 

Big Tech’s control over these four pillars of such power—information, finance, production, and security—are providing it with structural power that challenges even the durable monopoly power of electric utilities. This structural power stems not just from market power, but from a private firm’s investment power to shape the regulatory framework within which it functions. Unlike the monopoly power long enjoyed by electric utilities, Big Tech’s structural power is tied to their ability to shape the rules that affect them due to the government’s dependence on their economic investments, the trillions of dollars they anticipate spending on AI-related investments, their tight control over information related to data center development, and their impacts on reliability of the grid. After decades of sitting on the sidelines in energy spheres, Big Tech is poised to exert significant structural power on the stakeholders in the grid space. This in no way suggests that an analysis of structural power in the grid space involves a simple bilateral relationship between Big Tech and Big Utilities. 

The electric grid space begs for a deeper analysis of competing levels of structural power of Big Tech and Big Utilities vis-à-vis many other players. Local governments (primarily through zoning and land use commissions), state governments (primarily through public utility commissions), regional planners (primarily through regional grid operators), and the federal government (primarily through the Federal Energy Regulatory Commission) all interact with each other and these entities to maintain and develop the U.S. grid. Many other actors in the grid space can exert degrees of structural power, including various other government actors; quasi-governmental entities; and various business interests surrounding the grid, including private equity players and critical supply chain providers. But this work attempts to first disentangle the respective power of Big Tech and Big Utilities and hopes to spearhead future analysis of the power dynamics surrounding these other actors in the grid space.

This short contribution here demonstrates why data centers, and particularly the largest called hyperscale facilities that are measured in gigawatts and acres, provide a master class in political economy. Although there are many dimensions, this highlights the risks of three dimensions: (1) asymmetries between costs and benefits, (2) regulatory capture, and (3) private control of public resources.

Second, Big Tech’s structural power is allowing them to influence government policy to serve their own growth. Big Tech has won the narrative, convincing the Trump Administration and other followers that AI dominance is a national security imperative and a matter of geopolitical survival. We are waiting on FERC to issue an order in June that may upend the entire federalism bargain that has existed in this space for the last century to fast-track interconnection of data centers to the public electric grid.[4] But Big Tech’s regulatory capture extends to state governments as well, where they have been able to negotiate extremely favorable tax breaks. This leaves local governments with little political power to resist such data center development. One need only look at NextEra Energy’s pending $67 billion agreement to acquire Dominion Power to see elements of capture at play.[5] Utilities are notoriously territorial, enjoying state-sanctioned monopolies over the customers within their serve areas. These are two of the largest utilities in the country, with Dominion’s territory including the largest concentration of data centers in the world. This agreement needs multiple layers of federal and state approvals to proceed, and reports are already circulating about the aggressive lobbying of the parties, possible rate increases for ratepayers in these areas, and concerns about the amount of political control should this merger be approved.[6] 

Third, the legal structure of utility ratemaking allows Big Tech and Big Utilities to create a double dose of structural power that can leave ratepayers like you and me subsidizing their activities. As I have articulated elsewhere to illustrate how Professor Strange’s pillars of production and finance interact, there are at least three likely scenarios that reflect the power dynamics of Big Tech and Big Utilities. They are not mutually exclusive, and it is likely we will see a mixture of these arrangements depending on the political appetite and financial analyses of a particular area. But in simple terms, Big Tech will either be (1) Big Utility’s best customers; (2) Big Utility’s biggest competitors; or (3) Big Utility’s biggest disappointment. 

In the customer scenario, where data centers brave the long lines to interconnect to the public grid, utilities profit from both the enormous capital costs to serve these “large load” customers as well as from the large amount of electricity usage. One need only look to Meta’s plans to build the Hyperion hyperscale center in rural Louisiana as an example. The Louisiana Public Service Commission approved a settlement agreement between Meta’s subsidiary and Big Utility Entergy that approved Meta directly financing seven new natural-gas power plants totaling more than 5.2 gigawatts, along with roughly 240 miles of 500-kV transmission lines and battery storage across multiple locations.[7] A five-gigawatt campus is enormous, requiring more electricity than that of many cities. These facilities will be built by the Big Utility, financed by Big Tech, but included as “system resources” that are connected to the grid in a way that could allow Entergy to include them in their rate base used to calculate rates for residential customers like you and me. This is just as Professor Strange suggested, with private firms shaping the very infrastructure within which others must operate. 

In the competitor scenario, where data centers enter into agreements with electricity generators directly, this can sometimes displace generating resources from otherwise serving the public grid. In a world with finite resources, this can reduce the amount of generation available to the public grid, increasing prices for all of us as our public utilities scramble to procure sufficient generation. This can also happen in co-located arrangements, like when Amazon Web Services set up shop right next to the Susquehanna Nuclear Power plant in Pennsylvania, contracting a portion of the previous public supply. 

In the last scenario, Big Tech will forgo the public grid entirely. This scenario goes by many names today, but proposals for a microgrid, behind-the-meter generation, or BYOP (Bring Your Own Power) may all be referring to data centers that self-supply. These data centers will not be connected to the public grid, meaning Big Utilities will be disappointed to lose revenue from capital investments and from electricity sales. As of today, Big Tech suggests this is not the preferred option, instead, wanting the reliability and redundancy benefits of an interconnected grid. But as pressures mount, some like the Project Jupiter hyperscale facility approved in New Mexico, is being proposed as a microgrid that will be served by fuel cells that run primarily on natural gas. At the very least, these scenarios may provide a lifeline for data centers while they wait to interconnect to the public grid. But it is possible that these data centers remain disconnected, creating a shadow grid that runs outside of the public one. Such a shadow grid would be unregulated to the same extent as other resources and insulated from the public interest obligations that have governed electricity infrastructure for the past century. The extent to which Big Tech chooses how to power its data center campuses will shape the grid and its relationships with Big Utilities around the country. Not to be lost in this taxonomy is the last line which bears repeating – it is Big Tech who holds the structural power and the leverage to make these determinations and obtain necessary approvals.

Together, Big Tech’s influence to shape the resources, providers, and locations of grid infrastructure, as well as the development of new regulatory tools, arrangements, and innovative rate structures, defines the next phase in electricity history that I have coined the “Tech Energy Transition.” The positive and negative impacts of this next phase will not be felt equally across the nation. The United States’ grid is a complex, century-old machine that is in desperate need of investment, reform, and transformation. Lawmakers and regulators have cobbled together rules and markets that try to balance economic efficiencies with reliability and the public interest in ways that allow most Americans to remain functionally disengaged from the intricacies of the electric grid. 

If we are not careful, we may cede control over what types of energy sources are going to power our grid, where they will be located, and which areas of the country bear the risks and rewards of such investments. Worse yet, we could lose control over public resources to power private profit ambitions. 

But data center development, public engagement, and its associated constraints can trigger innovations that might not have otherwise been tenable. The grid may see a much-needed infusion of cash and regulatory creativity. By harnessing data center demand, regulators can guide the grid’s transformation in ways that are compatible with public policy goals like decarbonization and ratepayer affordability in ways to guide stakeholders during this Tech Energy Transition. 

If you want to learn more, please keep an eye out for my forthcoming book with MIT Press, The Data Center Next Door (Spring 2027). It is an essential portal for academics and non-academics alike to better understand the complicated tradeoffs associated with the energy, environmental, technological, and policy implications of data center development.

Amy L. Stein is Cone Wagner Professor of Law, University of Florida Research Foundation Professor. You can reach her by e-mail at stein@law.ufl.edu.