The Atlas connects data center infrastructure to the political and planning processes that shape its development. It allows users to trace relationships among facilities, policies, political events, and the actors involved in building, governing, or contesting them. Users can investigate individual projects and community conflicts, compare policy responses across jurisdictions, trace an operator’s development footprint, or examine broader patterns in how data center growth affects land, resources, public finance, and local governance.

The interface supports exploration across geographic scales and linked datasets. Users can search and filter records, move from places or operators to related activity, and follow connections across the Data Centers, Events, and Policy layers. The Details window brings these relationships together with supporting sources and structured exports for further research and analysis.

The Atlas is designed for researchers, students, community organizations, journalists, planners, policymakers, and others seeking to understand how the rapid expansion of data center infrastructure is reshaping places and generating new forms of political response.

Sources and inclusion

Records are built from publicly accessible evidence, including government and planning documents, legislative and regulatory records, company materials, journalism, research publications, and community documentation. Sources are evaluated according to the claims they can support. Official or primary records generally take precedence when they directly establish a project, decision, policy action, or other fact, while journalism and other secondary sources are often important for documenting changing project plans, disputes, and political activity. Community sources are used to document positions and participation rather than treated as neutral evidence for contested factual claims.

Discovery is separate from publication. News searches, public maps, legislative databases, and other sources are used to identify possible records as leads, but these enter the published Atlas only after review of their identity, relevance, geography, attributes, sources, and relationships. Published records remain grounded in the underlying public evidence. Large language models and other automated research tools may assist with discovery, translation, summarization, classification, and reconciliation, but they are not treated as sources. The workflow below shows how agent-assisted research is separated from human review and publication.

Research that produces no material change is logged so that completed checks, source sets, and unresolved questions can inform later review. Records that move through this workflow are then structured according to the following conventions.

Constructing records and relationships

For data centers, the main unit is the physical campus or site rather than an individual building, tenant deployment, permit, or phase announcement. Records distinguish current lifecycle status from development stage and use broad capacity categories where exact and comparable megawatt figures cannot be established. Locations represent the most precise defensible WGS84 point available, ranging from verified parcels or addresses to broader geographic estimates. Positional confidence describes the precision of that location rather than the overall reliability of the record.

Events capture discrete political, regulatory, legal, planning, environmental, or community developments. Policies represent formal governmental instruments that materially address data centers or their associated infrastructure and resource demands. Actors provide stable identities for organizations. Relationships are added only where sources support a connection at the relevant scale, such as direct involvement in a campus, corporate hierarchy, a durable institutional relationship, or participation in a specific event. Proximity, shared geography, or thematic similarity may suggest avenues for investigation but do not by themselves establish a formal relationship.

Validation, updates, and interpretation

The Atlas combines agent-assisted research and reconciliation, deterministic validation, and substantive human review. Automated checks test identifiers, dates, coordinates, classifications, lifecycle combinations, relationships, and likely duplicates, while human review evaluates proposed changes that require judgment about record identity, interpretation, or structure. Records are updated as projects and policies change, but this can be periodic and gaps may persist as data is updated weekly or monthly. Stable Atlas identifiers are retained wherever the underlying object remains the same. Public releases are tied to defined dataset revisions so that changes can be tracked over time.

Coverage is broad but uneven. Public documentation varies substantially across places, and U.S. legislative coverage is currently more systematic than many international and non-bill policy domains. Absence from the Atlas should therefore not be interpreted as evidence that a facility, controversy, policy, or institutional relationship does not exist. Capacity and location precision also vary across records. More generally, the Atlas documents observable sequences, associations, and evidence-supported relationships; it does not aim to establish causation. Researchers should therefore carry source quality, missing data, geographic precision, and collection bias into subsequent spatial, statistical, comparative, or qualitative analysis.

Data sources

Map layers and geography

European Commission, Joint Research Centre. "Global Forest Cover 2020 (GFC2020), Version 4." Forest Observations, 2026. https://forobs.jrc.ec.europa.eu/GFC. (Forest layer)

Natural Earth. "1:10m Cultural Vectors: Admin 0 – Countries; Admin 1 – States, Provinces; Admin 2 – Counties." Natural Earth, n.d. https://www.naturalearthdata.com/downloads/10m-cultural-vectors/. (National, subnational, and US county boundaries)

OpenStreetMap contributors. "OpenStreetMap." OpenStreetMap Foundation. Accessed September 18, 2026. https://www.openstreetmap.org/. (Base map, via Protomaps, and the power line and substation layers. Licensed ODbL)

Potapov, Peter, Svetlana Turubanova, Matthew C. Hansen, Alexandra Tyukavina, Viviana Zalles, Ahmad Khan, Xiao-Peng Song, Amy Pickens, Quan Shen, and Jocelyn Cortez. "Global Maps of Cropland Extent and Change Show Accelerated Cropland Expansion in the Twenty-First Century." Nature Food 3 (2022): 19–28. https://doi.org/10.1038/s43016-021-00429-z. Data: Global Land Analysis and Discovery (GLAD), University of Maryland. https://glad.umd.edu/dataset/croplands. (Cropland layer, 2019)

Protomaps. "Protomaps Basemap." Protomaps. https://protomaps.com/. (Tiles for the base map context archives)

TeleGeography. "Submarine Cable Map." TeleGeography, 2026. https://www.submarinecablemap.com/. (Cables and landingpoints geometry)

Records and relationships

Global Legal Entity Identifier Foundation (GLEIF). "LEI Data and Level 2 Relationship Data." GLEIF. Accessed September 18, 2026. https://www.gleif.org/en/lei-data/gleif-golden-copy.(Operator and actor ownership network)

LegiScan. "LegiScan API: State and Federal Legislation." LegiScan LLC. Accessed September 18, 2026. https://legiscan.com/legiscan. (US bill status, sponsors and actions)

Citing this work

The data center, event, and legislation records come from the project's own research workbooks, compiled from many primary sources included in each record and is publicly available. For citing this work, please use:

Kollar, Justin. Atlas of Data Center Politics. Version 0.1.0. Infrastructure Lab, National University of Singapore, 2026. https://www.atlasofdatacenterpolitics.com/. Licensed under CC BY 4.0.