Social Network Analysis of Texas Data Center Infrastructure Ownership

By: Julia Chialastri

---

Executive Summary & Key Findings

Dataset Scope: Analyzes 517 operational and planned data center facilities across

Texas, utilizing Social Network Analysis (SNA) to map relationships beyond simple

geographic location counts.

Ownership: While direct facility records indicate 96% U.S. ownership, multi-hop SNA

tracing reveals that 10% of total facilities are ultimately controlled by foreign parent

entities.

Local Market Resiliency: Texas-headquartered entities directly own 46% of all U.S.-

owned facilities in the state.

Structural Sparsity: An extremely low network density (0.001) and an average connection degree of 1.055 illustrates that Texas market ownership is fragmented.

Introduction

Modern corporate ownership is rarely linear. Between subsidiary layers, joint ventures, holding groups, and institutional investments, understanding who ultimately owns and operates critical economic infrastructure has become a remarkably complex task. [1] [2] [3] Traditional business intelligence relies heavily on multiple spreadsheets to record customers, competitors and sales prospects to trace organizational relationships.

Social Network Analysis (SNA) offers a transformative solution to this challenge. By shifting the perspective from isolated spreadsheets to network graph models, SNA automates and clarifies the laborious process of mapping entity-to-entity relationships, connecting parent companies to subsidiaries, and untangling complex ownership structures.[1] [2]Rather than treating organizations as discrete entries on a flat row, SNA frames them as nodes connected by meaningful transactional, governing, or financial relationships (edges).[3] [4] This structural approach elevates standard business intelligence into a predictive, multi-hop analytical framework.

To demonstrate the real-world power of SNA in clarifying opaque markets, this study examines the Texas data center infrastructure market. Earlier efforts this year to provide a count of the number of data centers in Texas have already become outdated due to the number of new projects and expanded details of known projects.  The data center sector serves as an ideal case study: it is a highly fragmented industry characterized by deep-rooted legacy operators, a rapid influx of capital-backed new entrants, and intricate leasing and ownership arrangements. By applying SNA to this sector, we move beyond simple geographic counts to map the true underlying architecture of ownership, offering policymakers, analysts, and industry leaders an unprecedented level of market transparency.

Research Methodology & Data Scope

While originally developed in sociology to map social interactions, Social Network Analysis (SNA) translates seamlessly to enterprise corporate structures by converting organizational ties into measurable networks.[1] In this framework, nodes represent physical data centers, operating entities, parent companies, and institutional investors, while edges quantify formal relationships, ownership stakes, lease agreements, and operational dependencies. Because spreadsheets remain the standard medium for corporate intelligence, SNA serves as a direct extension of existing workflows. Standard network analysis tools like Gephi import spreadsheet-formatted data directly, transforming static tables into mathematically measurable, multi-hop relationship graphs.[2]

Scope and Verification

Mapping data centers involves two primary accounting hurdles: non-standard naming conventions across public filings (building permits, tax exemptions, ERCOT interconnection requests, colocation listings) and ambiguity surrounding campus versus individual building counts. To ensure consistency, this dataset counts individual operational buildings rather than entire multi-building campuses, while excluding small-scale server rooms and hobby server set-ups. Under these criteria, the study identifies 517 distinct data center facilities across Texas.

To resolve project codenames and remove duplicate entries, a multi-step verification process was executed. Precise street addresses were mapped for all verified operational facilities, whereas proposed developments lacking physical addresses were assigned county-seat coordinates as spatial proxies to capture regional clustering. To prevent false duplication at these proxy points, project names, developer identities, and ownership filings were cross-referenced. Overlapping coordinate clusters were manually reviewed using corporate portfolios, local news archives, and, as a last resort, high-resolution satellite imagery to verify distinct structures on-site.

Ownership mapping

Direct and global ultimate ownership was determined through primary public records, including SEC filings, building permits, and corporate websites, supplemented by business registries such as Dun & Bradstreet and Open Corporates. While traditional financial definitions strictly define global ultimate owners through formalized merger and acquisition filings, this study classifies majority shareholders as global ultimate owners. Because a majority shareholder holds governing control and a vested strategic interest in expansion, treating them as ultimate owners aligns directly with the goal of evaluating overall market control.

Results: Texas Data Center Ownership Landscape

The final SNA network contained 517 data centers. The SNA metrics, as well as discussion on ownership breakdowns, are located below the graph. The ownership network of Texas Data center is a sparse network. It is not a densely knit community, instead companies and facilities are loosely connected along long linear pathways, relying heavily on key central actors to bridge distinct clusters.

Figure 1. Geographic distribution of corporate headquarters network.  Orange nodes represent companies headquartered in Texas; white and grey nodes represent companies headquartered in other US states; purple nodes represent international headquarters. Graph created by author using Gephi.

Network-Level Metrics:

Metric	Value	Insight
Network Density	0.001	Indicates an extremely sparse, non-mesh network where corporate entities operate in isolated silos rather than cross-invested webs.
Average Degree	1.055	Quantifies a hub-and-spoke structure built using single-purpose LLCs connected upward to single managing entities.
Network Diameter	12	Reveals deep organizational layering, showing beneficial ownership can be buried up to 12 entity hops deep
Average Path Length	4.3	Demonstrates that finding relationships between related market assets requires navigating over 4 intermediate hops, proving why multi-hop graph analysis outperforms flat spreadsheets.

Ownership Location

• Direct Ownership Findings (N=517):

  • 96% (498 facilities): Directly owned by US entities.

    • Texas HQ: 229 facilities (46% US-owned).

    • Other US States: 269 facilities (54% US-owned).

  • 4% (19 facilities): Directly owned by non-US entities

    
    

• Global Ultimate Ownership Findings:

  • Shift in ownership when tracing parent entities: US ownership falls to 90% (465 facilities), while foreign global ultimate ownership rises to 10% (52 facilities).

  • Concentration: Ownership spans 157 distinct private companies, universities, and school districts, with the Top 5 entities controlling 26% of the total market by facility count.

Figure 2. Geographic distribution of Texas Data Center Location.  Southern Texas is Light Green, Northern Texas is blue, West Texas is Yellow, Panhandle is Forest Green, Coastal is Pink. Datacenters with unlisted addresses are colored white.  Graph created by author using Gephi.

Sub-Population Focus: Tax-Exempt Data Centers

Tax-exempt data centers were designated as a special population due to their disproportionate impact on municipal tax bases and local revenue. As of August 1, 2026, 147 Texas facilities received sales tax exemptions. According to The Texas Tribune, these tax breaks represent an estimated $3.2 billion in foregone state revenue over the next two years alone.[1]

  The network of tax-exempt data centers is exceptionally sparse and fragmented, with most facilities remaining isolated or loosely tethered to just one or two neighbors. Rather than forming a tightly knit community, nodes are stretched across long linear pathways where information or influence must cross an average of 4.28 steps (and up to 12 hops) within connected corporate ownership sub-clusters. Because connectivity is thin and spread out, communication relies heavily on a few central intermediaries to bridge distant, isolated clusters.

Figure 3. Network connections between commercial entities and data center infrastructure. Blue Nodes are Data Centers and Data Campuses, Orange Nodes, Pink Nodes are Government Backed Projects. Graph created by author using Gephi.

Subpopulation Network-Level Metrics

Metric	Value	Note
Total Count	142	Five unresolved entities are excluded from the sub-population metrics.
Network Density	0.005	Network is still sparse, even when limited to tax-exempt data centers.
Average Degree	1.19	Most nodes are connected to 1 to 2 nodes (roughly), this indicates a low overall connectivity. Most nodes are peripheral, or completely isolated.
Network Diameter	12	The longest shortest path between any pair of connected nodes is 12 hops, indicating that the network is spatial spread out.
Average Path Length	4.28	On average, any two connected nodes are separated by 4 steps.

Tax-Exempt Sub Population Ownership

• Global Ultimate Ownership: 110 (77.5%) US ultimate, and 32 (22.5%) foreign ultimates.

• Ownership Depth: Foreign-owned tax-exempt facilities traverse an average of 2 intermediary holding entities before reaching the true global ultimate owner.

Discussion

Although data centers have maintained a footprint in Texas since the 1990s, early developments were sparse and carried minimal grid impact. [1] [2]The sector’s evolution has changed from rooms in the 1970s to mid-sized buildings in the 1990s, to expanding square footage in the 2010’s to accommodate the buildout of cloud, to the recent artificial intelligence boom leading to extreme market fragmentation as the number of individual data centers rapidly increases. [3] [4] Despite the booming market, local ownership remains surprisingly resilient, with nearly half (46%) of all Texas data centers being owned by entities headquartered directly within the state.

While the prevailing narrative suggests that digital infrastructure is monopolized by a few hyper-visible tech giants, the current network analysis presents a view of a fragmented market. The market power spans a multi-layered web of private equity, private firms, telecom operators, real estate investors, universities, and international holding companies. Alongside legacy operators, this AI surge has attracted new entrants from other industries.

Current regulatory challenges around data centers are further compounded by transparency issues in local fiscal governance. The finding that foreign-owned, tax-exempt entities traverse an average of two intermediate holding companies underscores how standard tax registry data can obscure true corporate beneficiaries. As highlighted by corporate disclosure monitors such as Good Jobs First and broader economic development research, standard state and municipal databases frequently fail to track complex, multi-tiered corporate families.[1] [2] [3]By routing property assets through multiple intermediate LLCs, these administrative structures impede local tax assessors from identifying ultimate beneficial owners, complicating efforts to evaluate whether state economic incentives yield genuine local economic retention or merely subsidize offshore entities.[4] [5]

Future Research

While this study maps the corporate network architecture of Texas data centers, evaluating true infrastructure risk requires pairing spatial corporate graphs with physical utility footprints. Future research will focus on two core integrations.

First, integrating network graphs with ERCOT interconnection queue data could provide insight into potential multiple queue positions under isolated shell LLCs, while overlaying ownership onto transmission maps will identify shared grid vulnerabilities. [1] More industry engagement with the Public Utilities Commission of Texas’ Energy and Water Usage Survey for Data Centers and Virtual Currency Mining Facilities could help drive additional insights into water usage profiles, cooling technology types, and local water permits. [1] Aggregating site-level power and water usage up through corporate ownership edges will reveal which global ultimate entities command the largest cumulative resource footprint.

Advancing research depends on transparent data. Operators, developers, and tenants must actively complete the Public Utility Commission of Texas (PUCT) Energy and Water Use Survey. Widespread participation bridges the gap between ownership mapping and physical resource impacts, providing decisionmakers with the data required to safeguard Texas energy and water resilience.

Conclusion

As demonstrated across Texas’s 517 data center facilities, critical economic infrastructure is rarely controlled through direct, linear arrangements. By framing corporate ecosystems as dynamic graphs of nodes and edges, Social Network Analysis (SNA) penetrates organizational opacity, uncovering an average path length of over four intermediate hops and revealing foreign ultimate beneficial ownership that standard filings undercount by more than half.

These structural insights hold immediate relevance for infrastructure planners, energy grid managers, and policy leaders. In an era marked by rapid AI expansion, capital-backed speculative entrants, and rising grid demands, understanding who ultimately owns, operates, and capitalizes data center assets is no longer just financial inquiry; it is a matter of state economic strategy and infrastructure resilience. By moving beyond static counts to map the true underlying architecture of ownership, SNA is positioned to provide decisionmakers with the market transparency necessary to safeguard grid stability, evaluate tax incentive efficacy, and navigate complex national security interests.

Figure 4. Spatial distribution of data center facilities. Map created by author using Power BI.

Sources:

1. Bastian, Mathieu, Sebastien Heymann, and Mathieu Jacomy. "Gephi: An Open Source Software for Exploring and Manipulating Networks." Proceedings of the International AAAI Conference on Web and Social Media 3, no. 1 (2009): 361–62. https://doi.org/10.1609/icwsm.v3i1.13937.

   
   

2. Bayat, Seyed Alireza, et al. "Application of Social Network Analysis (SNA) in Business Intelligence and Entity-to-Entity Relationship Mapping." Working paper, 2025.

   
   

3. Cobler, Paul. 2026. "Texas Is Giving Data Centers More Than $1 Billion in Tax Breaks Each Year." The Texas Tribune, April 8, 2026. https://www.texastribune.org/2026/04/08/texas-data-centers-sales-tax-break-billion-dollars/.

   
   

4. Electric Reliability Council of Texas (ERCOT). System Planning & Large Flexible Load Task Force (LFLTF) Capacity Reports. Austin: ERCOT.

   
   

5. Ellis, S. "The Data Center Boom Is Reshaping Texas — and Every Business Needs to Understand Why." JD Supra, May 27, 2026. https://www.jdsupra.com/legalnews/the-data-center-boom-is-reshaping-texas-4072838/.

   
   

6. Fath, Brian D., et al. "Systemic Vulnerabilities in Networked Economic and Infrastructure Systems." Ecological Modelling 411 (2019): 108801.

   
   

7. Gaspar, J., and P. Le Billon. "Digital Infrastructure Geography." Ongoing research, 2022–.

   
   

8. Hoopes, Jeffrey, Leslie Robinson, and Joel Slemrod. "Corporate Tax Disclosure." NBER Working Paper No. 31467, National Bureau of Economic Research, 2023. https://doi.org/10.3386/w31467.

   
   

9. Jackson, Matthew O. "Representing and Measuring Networks." In Social and Economic Networks, 19–51. Princeton: Princeton University Press, 2008.

   
   

10. Jensen, Nathan M., and Calvin Thrall. "Who's Afraid of Sunlight? Explaining Opposition to Transparency in Economic Development." Business and Politics 23, no. 4 (2021): 474–91. https://doi.org/10.1017/bap.2021.8.

    
    

11. New America. "The Evolution of Data Centers." In Securing the Backbone of AI. Washington, DC: New America, September 2025. https://www.newamerica.org/insights/securing-the-backbone-of-ai/the-evolution-of-data-centers/.

    
    

12. OECD. Institutional Investor Engagement and Stewardship. Paris: OECD Publishing, 2025.

    
    

13. Public Utility Commission of Texas. "Energy and Water Use Survey for Data Center and Virtual Currency Mining Facilities FAQs." Accessed November 9, 2026.

    https://www.puc.texas.gov/industry/water/utilities/energy-and-water-use-survey/faq/.

    
    

14. Stream Data Centers. "Dallas Area Data Center Solutions." Accessed August 27, 2026. https://www.streamdatacenters.com/locations/dallas-data-centers/.

    
    

15. Tarczynska, Kasia. Even Cloudier with a Greater Loss of Spending Control: How Data Center Tax Abatements Undermine Public Budgets. Washington, DC: Good Jobs First, June 2026. https://goodjobsfirst.org/even-cloudier-with-a-greater-loss-of-spending-control-how-data-center-tax-abatements-undermine-public-budgets/.

    
    

16. Tierney, Susan F. "ERCOT Texas's Competitive Power Experience: A View from the Outside Looking In." White paper, Analysis Group, October 2008.

    
    

17. https://www.analysisgroup.com/globalassets/content/insights/publishing/tierney_ercot_texas_study_11-082.pdf.

    
    

18. Vitali, Stefania, James B. Glattfelder, and Stefano Battiston. "The Network of Global Corporate Control." PLoS ONE 6, no. 10 (2011): e25995. https://doi.org/10.1371/journal.pone.0025995.