Are Data Centers Just Another Industrial Asset?

Slide 1. Are Data Centers Just
Another Industrial
Asset?
Bringing some data to bear on a vital and politically charged commercial opportunity space.
Gabriel Collins, J.D.
26 August 2026

Slide 2. IMPORTANT NOTICE
Disclaimer & Disclosures
This presentation is provided for information and discussion purposes only. Consumers of the information and analysis contained in it make investments and take other actions at their own risk.
I am not a hydrologist, geologist, or engineer, and am not offering advice on technical aspects of any assets which may be discussed in this analysis. The information and opinions contained in, and expressed by, this analysis are based on sources I deem to be reliable. However, no warranty, assurance, or guarantee, express or implied, is made about the completeness, reliability, or accuracy of this content. Likewise, nothing in this presentation or the accompanying discussion creates an attorney-client relationship between us. I serve as a strategic advisor to Revive Water
Resources, LLC.
I have used Anthropic’s Claude and OpenAI’s ChatGPT to collect data and adversarially test my ideas and visually format the slides and Google’s Gemini to create some images.
The views expressed herein are my own interpretations as of the date of presentation, are made in my personal capacity, and are subject to change without notice.

Slide 3. ROADMAP
What We’ll Cover
I The Thundercloud Charges Up
Texas press clippings, January to August 2026: pause, moratorium, tax rollback, ban, audit.
II What Data Centers Actually Are — and Where They’re Going
The layers we argue about, the layers we forget, and Silicon Valley’s own 10,000× forecast.
III Are They Just Another Industrial Asset?
Benchmarking location, buildout velocity, and power and water against past industrial booms.
IV The Silicon South Plains
Why the Permian baseline is different — and the acreage and capital math behind a gigawatt.
V Making Them Better Neighbors
Water optimization, noise, sightlines, and thermal signature.

Slide 4. I
The Thundercloud Charges Up

Slide 5. TEXAS PRESS CLIPPINGS · JAN–AUG 2026
The Backlash Becomes Policy PAUSE → MORATORIUM → TAX ROLLBACK → BAN → AUDIT
THE TEXAS TRIBUNE / JUN 30, 2026
San Marcos becomes the first
Texas city to ban data centers, testing its local control
THE TEXAS TRIBUNE / FEB 13, 2026
Texans are demanding their local governments push pause on data centers. Can they?
HOUSTON CHRONICLE / MAY 20, 2026
Sid Miller calls for data center moratorium, deepening GOP divide over construction
THE DALLAS MORNING NEWS / MAY 19, 2026
North Texas officials split votes on future data center development
SAN ANTONIO EXPRESS -NEWS / JUN 4, 2026
Hill Country residents push water utility to say no to data center
THE TEXAS TRIBUNE / JUN 10, 2026
Abbott recommends sweeping data center regulation, including eliminating sales tax exemption
HOUSTON CHRONICLE / AUG 3, 2026
Greg Abbott orders sweeping audit of Texas data centers amid power grid concerns
7 VERIFIED HEADLINES
Reproduced as published.
Verified 3 August 2026.
Sources: The Texas Tribune; Houston Chronicle; The Dallas Morning News; San Antonio Express-News. Full article URLs are recorded in the speaker notes.

Slide 6. II
What Data Centers Actually Are — and Where They’re Going
The layers the public argues about, the layers everyone forgets, and the industry’s own forecast.
ONIONS · NOW VS. NEXT · 10,000×

Slide 7. THE PUBLIC DEBATE
Datacenters Are Like Onions
WE FOCUS ON THIS PART AND FORGET THIS PART
Framing after Gabriel Collins, “Datacenters Are Like Onions,” The Sinews of Civilization, 6 August 2026.

Slide 8. CAPABILITY TRAJECTORY
What Are Some Things Datacenters Do Now…And What’s Likely Coming Next?
NOW / NEXT
Frontier models in production, and inference already riding in the vehicle.
Next: discovery, diagnosis and materials science at machine speed.

Slide 9. SILICON VALLEY’S OWN FORECAST
Where Does Silicon Valley See AI Going?
“We conjecture that in five years 10,000x more AI inference will be done worldwide, with these gains hypothetically coming from multiplicative progress of 50x in AI algorithms, 50x in system/hardware optimization/specialization, and 4x from further data center growth.”
50×
AI ALGORITHMS
× 50×
SYSTEM / HARDWARE
× 4×
DATA CENTER GROWTH
= 10,000×
MORE AI INFERENCE BY 2031
— Computing Research Association Ideas Dinner, 27 April 2026 https://cra.org/industry/2026/04/27/how-ai-will-reshape-computer-systems-by2035-a-jeffersonian-dinner-in-san-francisco-about-our-10000x-future/ For those of us in the dirt and physical asset worlds, this is the domain of highest interest.

Slide 10. III
Are They Just Another
Industrial Asset?
Three benchmarks against the booms that built the last industrial map.
LOCATION · BUILDOUT VELOCITY · POWER AND WATER

Slide 11. BENCHMARK 02 · BUILDOUT VELOCITY
How Do Hyperscale Datacenters Compare to Other Industrial Assets?
STEEL PRODUCTION
CAGR, 1870–1944
RAILROAD MILEAGE
CAGR, 1840–1916
DATACENTER CAPACITY
CAGR, 2000–2025
▪ Booms don’t usually accelerate as they near their peak.
▪ But datacenters are doing just that—and we do not even know where the peak might be.
▪ Datacenter capacity shows big acceleration toward the back end of the curve, with capacity potentially doubling again between 2026 and 2030.
▪ The World War II steel surge relied heavily on capacity built 15–20 years prior that was underutilized during the Depression, then surged during World War II.
U.S. steel production, 1,000 short tons.
Methodology: starting point is a judgment call on when true “industrial scale” development commenced; endpoint is the all-time historical peak figure.
Sources: U.S. Census Bureau (Hist. Stat. 1975, 1949); NBER/FRED (2026); USGS (Data Series 140, 2024); AISI (Ann. Stat. Rep.); U.S. Dept. of State (1814); Swank (1881, 1902); Fishbein (NARA, 1973); Kennedy (1859); Allen (JEH, 1979).

Slide 12. BENCHMARK 01 · KEY METRIC: LOCATION
How Do Hyperscale Datacenters Compare to Other Past Industrial Boom Assets?
A “steel pig” pouring molten steel into a mold. Jones & Laughlin Steel Company, Pittsburgh, PA, 1942. U.S. National Archives and Records Administration.
Steel plant siting context: U.S. National Park Service, https://www.nps.gov/articles/000/pittsburgh-pennsylvania.htm

Slide 13. BENCHMARK 01 · KEY METRIC: LOCATION
Hyperscale Datacenters vs. Past Industrial Booms: Railroads
Source: Louis P. Cain, “Railroad Mileage and Equipment: 1830–1890,” tables Df874–881, and “Railroad Mileage, Equipment, and Passenger Traffic and Revenue: 1890–1980,” tables Df927–955, in Historical Statistics of the United States, Earliest Times to the Present:
Millennial Edition, ed. Susan B. Carter et al. (New York: Cambridge University Press, 2006), https://doi.org/10.1017/ISBN-9780511132971.Df865-1111; National Bureau of Economic Research, “Increases in Railroad Mileage for United States,” NBER Macrohistory
Database, series A02082USA374NNBR, via Federal Reserve Bank of St. Louis FRED, accessed July 31, 2026; U.S. Interstate Commerce Commission, Annual Report on the Statistics of Railways in the United States, annual issues (Washington, DC: U.S. Government Printing
Office, 1887–1954); and U.S. Census Bureau, “May 2023: The Transcontinental Railroad,” History Stories, May 2023.

Slide 14. THE PHYSICAL FOOTPRINT
Datacenters Still Cluster, But Are More Geographically Dispersed
Source: https://docs.nlr.gov/docs/gen/fy26/98020.jpg

Slide 15. BENCHMARK 03 · POWER AND WATER
How Do
Hyperscale
DCs Compare to Other
Industrial
Assets?
Industrial and infrastructure facilities: electricity load vs. water load, log scale.

Slide 16. IV
The Silicon South Plains
Why the Permian Basin is different — and what a gigawatt likely requires in acres and capital.
BASELINE · LEASING MATH · INFERENCE SITING

Slide 17. REGIONAL ADVANTAGE
The Silicon South Plains
▪ Advantaged gas pricing
▪ Advantaged solar buildout
▪ Space to work with
▪ Less NIMBY-ism when single surface owners’ lands are often measured in square miles
THE RIGHT BASELINE
The baseline for disruption is not “how does this datacenter change things in Hill
Country Resort X?” but rather, “How do things change compared to the baseline of an oilfield region that collectively produces more than every country in OPEC other than Saudi Arabia?”
Permian Basin FracFocus well density and data center sites.

Slide 18. LAND AND CAPITAL MATH
Datacenter Development: Leasing Implications
1–3 acres per megawatt of compute
Rule of thumb for the surface needed for the campus itself.
+25% for linear infrastructure
Added on top for inbound power lines, water lines, and similar.
1,250–4,000 acres per gigawatt
Land work needed to secure each gigawatt of compute.
$25 billion capex per gigawatt
A gigawatt of compute can equate to this much capital.
If the project has associated wind or solar generation, those acreage figures rise dramatically.

Slide 19. SITING FOOTPRINT
Inference Datacenters: Physically Smaller But Steep Growth Ramp
Inference hardware: NVIDIA Blackwell NVL72 rack.
55× site footprint ratio
Stargate Abilene (1,100 acres) vs. Project Infrared (20 acres).
Sources: https://www.reporternews.com/picture-gallery/news/2025/09/25/inside-abilenes-massive-1100-acre-500-billion-project-stargate-texas-newspaper/86332564007/; https://www.everythinglubbock.com/news/local-news/deal-between-leda-and-data-center-company-corvadiscovered/; https://cdn.wccftech.com/wp-content/uploads/2024/10/NVIDIA-Blackwell-AI-Foxconn-Server-NVL72-1920×1039.png

Slide 20. V
Making Them
Better Neighbors
The mitigation actions that can decide whether a campus is accepted or fought.
WATER · NOISE · SIGHTLINES · THERMAL

Slide 21. THE MITIGATION THESIS
Fewer Tears, Great Food
FEWER TEARS / GREAT FOOD
The point of mitigation is not to stop cutting onions. It is to get the meal without the tears.

Slide 22. MITIGATION · WATER
Water Optimization
A WORKING EXAMPLE / TEXAS CONTEXT: PRECIPITATION GRADIENT WILL AFFECT COOLING CHOICES
▪ Alibaba Data Centers in Ulanqab, China (40.98, 113.24)
▪ Most of year the facility uses air cooling
▪ During the warmest months, they use indirect evaporative cooling

Slide 23. MITIGATION · NOISE
Noise Mitigation
× THE LITIGATION PATH ✓ THE ENGINEERING PATH
NTT TX4 Datacenter, Garland, TX — sound wall panels slid into the existing steel structure.
Source: https://www.ailsoundwalls.com/project/ail-sound-wall-panels-slide-into-existing-steel-structure-to-quiet-ntt-data-center-in-texas/

Slide 24. MITIGATION · SIGHTLINES
Visual Mitigation

Slide 25. MITIGATION · THERMAL SIGNATURE
Thermal Mitigation
Source: https://www.dvidshub.net/image/6607975/1-25th-arb-welcomes-spring-with-aerial-gunnery-range

Slide 26. Thank You
Gabriel Collins, J.D.
Baker Institute for Public Policy– https://www.bakerinstitute.org/expert/gabriel-collins
The Sinews of Civilization– gabrielcollins.substack.com
26 August 2026

Slide 27. THE WRITTEN RECORD
Further Reading
▪ Collins, Gabriel. Could the Permian Basin Become America’s Next AI
Data Hub? Houston: Rice University’s Baker Institute for Public Policy, August 15, 2025. https://doi.org/10.25613/CDF1-4006.
▪ Collins, Gabriel. “Recycled Water for Texas Data Centers: A Model
Supply Agreement.” Texas Water Intelligence, December 12, 2025. https://texaswaterintelligence.com/2025/12/12/recycled-water-fortexas-data-centers-a-model-supply-agreement/.
▪ Collins, Gabriel, and Christopher Bronk. AI Geoeconomics: Allied
Electricity and Secure Data Centers. Houston: Rice University’s Baker
Institute for Public Policy, December 18, 2025. https://doi.org/10.25613/T8RH-7H63.
▪ Collins, Gabriel. “In the Permian, Oil and AI Will Compete for Electricity.” The Sinews of Civilization: Fire, Food, Water, Force.
Substack, February 13, 2026. https://gabrielcollins.substack.com/p/inthe-permian-oil-and-ai-will-compete.
▪ Collins, Gabriel. “Texas AI Data Centers: Electric Grid Capacity Is a Bigger Challenge Than Water.” The Sinews of Civilization. Substack, February 19, 2026. https://gabrielcollins.substack.com/p/the-realbottleneck-for-ai-infrastructure.
▪ Collins, Gabriel. “When AI Data Centers Become Drone and Missile
Targets.” The Sinews of Civilization. Substack, March 2, 2026. https://gabrielcollins.substack.com/p/when-ai-data-centers-becomedrone.
▪ Collins, Gabriel. “AI vs. Heavy Industry: How Do Datacenter Electricity and Water Loads Actually Compare?” The Sinews of Civilization.
Substack, May 13, 2026. https://gabrielcollins.substack.com/p/aidatacenter-electricity-water-heavy-industry.
▪ Collins, Gabriel. “The Data Center Industry Needs Its Own FracFocusStyle Transparency Initiative.” The Sinews of Civilization. Substack, July
2, 2026. https://gabrielcollins.substack.com/p/data-centers-fracfocustransparency-computefocus.
▪ Collins, Gabriel. “Datacenters Are Like Onions.” The Sinews of Civilization. Substack, August 6, 2026. https://gabrielcollins.substack.com/p/ai-datacenters-power-waterpolicy-solutions.