By Javier Surasky
.png)
Data centers are entering the U.S. electoral debate, and I argue that this is a sign of a reality that, sooner rather than later, we will see exposed elsewhere, because behind the disputes over electricity, water, and land lies a question that goes beyond the United States: Who decides what costs communities must bear to sustain global technological competition?
The race
for artificial intelligence has finally brought into view a debate that
directly affects people’s pockets: Who pays the bill for staying in the
competition?
In the
United States, that question has reached the ballot box and, according to the
survey published by Ballotpedia in September 2026, 34 local initiatives related
to data centers will be put to a vote on November 3, alongside the U.S. midterm
elections. They are part of the 38 measures identified for all of 2026, twenty
of them located in Ohio, and include bans, moratoriums, and land-use decisions.
Although
not all the November ballot questions on data centers will be binding, they
share the opening of the electoral space to decide on this AI infrastructure.
On October
4, Donald Trump was in Ohio and defended data centers during a campaign event
by linking the necessary investments to competition with China, explaining that
rejecting them could benefit his main competitor. In doing so, he sought to
translate a local dispute into the language of rivalry between great powers
(Associated Press, 2026).
For a
community discussing what will happen to its resources, its territory, or its
electricity bill, Trump’s argument becomes a claim that local communities
should commit themselves to the global technology race, even by accompanying
decisions made without their participation by those building AI infrastructure.
The digital future needs somewhere to be installed
We have
already noted in previous posts that the expansion of AI depends on physical facilities connected to electrical systems and located in specific territories,
that the “cloud” is very concrete, and that it coexists with other users with
whom it competes for the resources it consumes.
Along these
lines, Young (2026) proposes understanding data centers as critical
infrastructure located at the intersection of public and private interests,
since, although they are privately owned and managed assets, their demands for
electricity, water, and land produce localized effects on specific populations
and public institutions.
Water is
part of that problem: depending on facility design and local conditions, data
centers can add pressure on already disputed water resources and, although not
all projects have the same consumption profile, they all need it in large
quantities.
Now, data
centers support many digital services beyond AI, and presenting any conflict
over them as a protest against artificial intelligence is misleading. Still,
there is no doubt that the growth of AI represents a relevant pressure: the
International Energy Agency estimates that data center electricity consumption
increased by 17% in 2025 and, in data centers exclusively dedicated to AI, the
increase reached 50% (International Energy Agency, 2026).
That
difference in scale helps us understand the conflict, because an investment may
be presented as a benefit for the national economy, while its most immediate
effects are concentrated in a locality that bears the material weight and
becomes the stage where promises of technological leadership and tensions over
resource availability converge.
Asking who
pays is not a rejection of technological development, just as opposing the
installation of a data center is not. Rather, it is a way of approaching how
its benefits, obligations, and risks are distributed, in relation to access to
adequate information, regulatory capacity, and authority to negotiate. This
brings us back to Young (2026), who identifies information asymmetries,
institutional fragmentation, and differences in technical capacity between
developers and local governments as part of the governance problem.
Seen this
way, the dispute is as much about the resources that infrastructure consumes as
it is about the capacity of governments and communities to negotiate the
conditions under which that infrastructure is installed.
The bill depends on political decisions
The
International Energy Agency ([International Energy Agency], 2026) notes that
the installation of a data center does not necessarily increase household
energy rates, since other variables intervene, such as the electricity system
and the policies adopted: where there is surplus energy capacity, additional
demand can make better use of infrastructure, but where supply is tight, it may
require new investments and spark friction over the distribution of a scarce
good.
Ohio offers
a clear example of this: in July 2025, the state’s Public Utilities Commission
ordered AEP Ohio, the electric utility that supplies much of its territory,
to establish specific tariffs for data centers after identifying the risk that
the investments needed to meet the enormous electricity demand expected from
these facilities would end up underused and that their costs would ultimately
be passed on to local users. The aim, then, was for large data centers to
assume a greater share of the economic risk associated with the infrastructure
needed to supply them (Public Utilities Commission of Ohio, 2025).
Recent
literature suggests that financing conditions can affect the social legitimacy
of projects: in an experimental study conducted in Germany, Heering and Voeten
(2026) found that projects involving significant increases in local electricity
prices received less support, as did those that placed high pressure on water
supply. While those results cannot be generalized—there is no available
evidence to do so—they provide a demonstration that, for communities, the
balance between technological benefits and local costs is not an abstract
question.
When a global dispute divides the territory
The
communities that coexist within a territory are heterogeneous actors, and the
installation of large-scale infrastructure can benefit some sectors, impose
costs on others, and open internal disputes over which development model should
be adopted.
Clinton, a
city in the state of Iowa, offers a clear example: the company QTS is planning
a hyperscale data center campus there whose investment could reach USD 10
billion and is supported by a local organization, Grow Clinton, which sees that
investment as an opportunity for growth. But a group of residents who
participated in the process of drafting the municipal ordinance called for
stricter controls on water, noise, air quality, and environmental monitoring
before moving forward (Billingham, 2026).
That
conflict does not pit an external company against a locality, but neighbors
against neighbors, transforming the question of who pays the energy and water
bills into the question of who can define the limit of what is acceptable
for the community.
Ashville,
Ohio, shows a different face of the problem. There, the municipal government
had moved forward with a cooperation agreement to facilitate the construction
of two data centers and a natural gas power plant, but a group of residents
gathered signatures to submit the decision to a referendum. The municipality
refused to certify the petition, which brought the controversy to the Supreme
Court of Ohio, which ordered that it be sent to the board of elections to allow
the vote (Trevas, 2026). In this second case, what emerges is the dispute over who
has the authority to decide on a large-scale territorial transformation.
Loudoun
County, Virginia, adds a third dimension: after decades of data center
expansion, the issue began to divide the local political leadership to the
point that two Democratic figures adopted different positions toward new
facilities: one pushed for restrictions on growth, while another defended the
industry’s fiscal contribution, and neighborhood opposition to new facilities
produced alliances that cut across traditional party preferences (Neuman,
2026).
The three
cases are different expressions of the same reality: Clinton shows a dispute
over the local distribution of costs and benefits; Ashville, a controversy over
participation and political authority; Loudoun, a reconfiguration of partisan
and social alignments.
That is
enough to reaffirm that large digital infrastructures do not land on passive
territories and can trigger debates over development, representation, and
distributive justice, create new coalitions, and turn decisions that were
previously seen as “merely administrative” into sources of conflict.
Geopolitics in the municipality
Trump’s
defense exposes an additional tension: international competition can be used as
an argument to accelerate territorial decisions, something that deserves to be
discussed.
Recognizing
the strategic value of computing capacity does not resolve which project should
be approved or under what conditions, nor does it determine how the burden of
creating that capacity should be divided among people and communities, or what
guarantees are owed to those who live near a planned new data center.
Heering and
Voeten (2026) found that, among the participants in their study in Germany,
explicitly emphasizing digital sovereignty produced only a marginal increase in
overall support for building data centers, while the operator’s identity had a
greater effect: participants showed preferences for German or European
operators over U.S. or Chinese companies. Sovereignty, therefore, matters.
The argument Trump advanced in Ohio and the work of Heering and Voeten show how the geopolitics of AI can be a central element of domestic distributive politics, and vice versa. From Global Radar, this invites us to broaden the view of AI power to bring onto the board a piece that represents the institutional capacity to agree on where infrastructure is installed and under what rules.
Social
acceptance does not necessarily depend on being for or against technological
expansion, but on more tangible elements for citizens, such as the financial
costs they will have to assume, potential benefits, environmental risks, noise
pollution, and several other factors that become part of a list that changes
from place to place.
A brake on the race or a dispute over its conditions?
I cannot
claim that electoral resistance will stop growth. What is more, Reuters (2026)
reported on October 5 that Goldman Sachs considers its immediate impact limited
and maintains its outlook for strong expansion of U.S. data centers through
2027.
Nevertheless,
the votes will allow us to observe something more precise: where there is
enough opposition to change decisions, what conditions voters are demanding,
and whether governments can make their technological ambitions compatible with
acceptable commitments for the territories that must host the infrastructure.
The ballot
box does not pose a dilemma between technological development and rejection of
infrastructure, but rather a dispute over the rules of that development:
planning, transparency, cost distribution, resource use, transparency, and the
possibility of local participation in decision-making.
For
countries seeking to attract digital infrastructure, the case raises a variable
they should pay attention to: hosting computing capacity can open
opportunities, but it can also be a source of social conflict. The announced
amount of an investment does not include that assessment.
At the U.S. ballot box, discussion is beginning over who has the authority to set the conditions, who participates in that decision, and how the costs of an infrastructure presented as necessary to compete on a global scale are distributed, and gaining positions against an international rival may be a government goal, but it is not necessarily among the priorities of the communities that must provide the territorial base and resources for the operation of the infrastructure needed to be at the forefront of the competition.
References
Associated
Press. (2026, October 4). Trump defends data centers in Ohio as he rallies
to boost Jon Husted in a tight Senate race. https://apnews.com/article/a86aad6b19d4afb2c505b902171ebd8e
Ballotpedia.
(2026, September 21). Ohio leads states in number of local ballot measures
on data centers for 2026. https://news.ballotpedia.org/2026/09/21/ohio-leads-states-in-number-of-local-ballot-measures-on-data-centers-for-2026/
Billingham,
E. (2026, July 16). Clinton residents request steep restrictions in the face
of QTS hyperscale data center. Iowa Public Radio. https://www.iowapublicradio.org/ipr-news/2026-07-16/clinton-qts-hyperscale-data-center
Heering,
J., & Voeten, E. (2026). How sovereign control, decarbonization and energy
costs shape public support for data centers. Nature Communications, 17,
9576. https://doi.org/10.1038/s41467-026-76500-9
International
Energy Agency. (2026). Key questions on energy and AI: Executive summary.
https://www.iea.org/reports/key-questions-on-energy-and-ai/executive-summary
Neuman, S.
(2026, September 21). The data center backlash is reshaping American
politics—one community at a time. NPR. https://www.opb.org/article/2026/09/21/data-center-backlash/
Public
Utilities Commission of Ohio. (2025, July 9). PUCO orders AEP Ohio to create
data center specific tariff. https://content.govdelivery.com/accounts/OHPUC/bulletins/3e8bb79
Reuters.
(2026, October 5). Goldman sees US data center growth intact despite
opposition. https://www.reuters.com/business/goldman-sees-us-data-center-growth-intact-despite-opposition-2026-10-05/
Trevas, D.
(2026, August 7). Village must submit data center referendum to county board
of elections. Court News Ohio. https://www.courtnewsohio.gov/cases/2026/SCO/0807/260906.asp
Young, C.
(2026). Clouds on the horizon: An integrative review of data centers and local
governance in the United States. Oxford Open Energy, 5, oiag006. https://doi.org/10.1093/ooenergy/oiag006