The Power Behind AI: Energy Demands + Utility Infrastructure
Opportunity on an almighty scale
The Power Behind AI: Energy, Infrastructure, and Real Estate Opportunities
Artificial intelligence (AI) is reshaping industries at an unprecedented pace, offering transformative capabilities in healthcare, finance, logistics, and countless other fields. Yet behind the seamless operation of AI-powered tools lies a hidden cost—energy. AI demands massive amounts of electricity and water to train models, process data, and maintain data center operations. This energy requirement is driving the need for specialized real estate and infrastructure to support AI's growth.
In this blog post, we’ll explore the energy and utility demands of AI, the industrial real estate that houses this infrastructure, and where future development opportunities lie for savvy investors and developers.
The Massive Energy and Water Requirements of AI
Energy Usage of AI
The power requirements for AI are staggering. Training a single large AI model, such as GPT, can consume as much electricity as 100 U.S. households use in a year. A single query to an advanced AI system may seem minor, but when scaled to millions of daily users, the energy demand becomes exponential.
Key drivers of AI energy consumption include:
Model Training: High-performance computing (HPC) systems train AI models by processing vast amounts of data over weeks or months.
Data Center Operations: The infrastructure supporting AI—including servers, cooling systems, and storage—requires continuous power.
Inference Tasks: Once trained, models require power to process user queries in real time, further contributing to electricity usage.
Water Usage for Cooling
In addition to electricity, AI operations rely heavily on water. Data centers generate heat during operation, and cooling systems often use water to regulate temperatures. Estimates suggest that an average data center consumes hundreds of thousands of gallons of water daily, with some hyperscale facilities requiring over a million gallons.
Utility Infrastructure: The Backbone of AI Operations
The infrastructure needed to support AI isn’t just about plugging into the grid. It requires robust and specialized systems to deliver reliable power and water.
Electricity Infrastructure
High-Capacity Power Lines: AI facilities often require access to high-voltage transmission lines to meet their energy needs.
Redundant Power Systems: Backup generators and battery storage are essential to ensure uninterrupted operations during outages.
Renewable Energy Integration: With sustainability goals in focus, AI facilities are increasingly powered by renewable energy sources, such as wind, solar, and even nuclear.
Water Systems
Water is crucial for cooling, and data centers are often located near abundant water sources. Advanced cooling technologies, such as liquid immersion cooling and closed-loop systems, are being adopted to reduce water usage, but the demand remains significant.
Connectivity and Fiber Infrastructure
AI data centers require ultra-fast connectivity to process and transmit data efficiently. Fiber optic networks, low-latency internet connections, and proximity to major network hubs are critical considerations.
Where Are AI Data Centers Located?
Existing Hubs
AI data centers and HPC facilities are concentrated in areas that offer:
Abundant Utilities: Reliable electricity, water, and network connectivity.
Proximity to Urban Areas: Close enough to population centers to reduce latency but far enough to avoid high real estate costs.
Economic Incentives: Tax breaks, subsidies, and favorable zoning laws attract developers.
Major hubs include:
Northern Virginia: Known as "Data Center Alley," this region hosts over 70% of the world’s internet traffic and benefits from abundant power and fiber connectivity.
Central Oregon: Offers low-cost hydroelectric power and cool climates that reduce cooling costs.
Phoenix, Arizona: A hotspot for hyperscale data centers, with solar energy integration and strategic access to fiber networks.
Dallas-Fort Worth, Texas: An affordable location with strong utility infrastructure and a central geographic position for U.S. data distribution.
Emerging Markets
As traditional hubs reach capacity, new markets are emerging.
Salt Lake City, Utah: Favorable climate, low utility costs, and increasing fiber connectivity.
Columbus, Ohio: Strategic location for east-west data routing and access to renewable energy.
Northern Sweden and Norway: Abundant hydropower and naturally cool climates make these regions attractive for sustainable data centers.
Real Estate Opportunities: Developing for AI Infrastructure
The growing demand for AI presents significant opportunities in real estate development. Here’s what developers and investors should consider:
1. Site Selection Criteria
Utility Availability: Sites must have access to high-capacity electricity and water infrastructure.
Proximity to Renewable Energy: Locations near solar, wind, or hydroelectric power sources will attract long-term investments.
Connectivity: Fiber optic infrastructure is critical for ensuring low-latency connections.
Zoning and Incentives: Favorable zoning regulations and economic incentives can reduce development costs.
-
Utility Availability
To support AI data centers, sites must have access to high-capacity electricity and water infrastructure. But what does "high capacity" mean in practical terms?
Electricity Requirements: High-capacity electricity is far beyond the capacity of a typical commercial building. Here’s a breakdown:
Amps: Small commercial buildings may use 400-800 amps, but large-scale AI facilities often require tens of thousands of amps. For example, a facility consuming 10 megawatts (MW) of power could need over 13,000 amps at 480 volts in a three-phase power configuration.
Three-Phase Power: This is essential for industrial-scale operations, offering consistent and efficient power delivery. Three-phase systems distribute electrical load more evenly, reducing strain on infrastructure and supporting high-demand equipment.
Megawatts: Hyperscale data centers demand significant power, typically ranging from 10 MW for smaller facilities to over 100 MW for the largest installations. For perspective, 1 MW can power roughly 750 average U.S. homes.
Infrastructure vs. Internal Distribution:
Infrastructure: Electricity must first be available at the site. This involves high-capacity transformers and grid connections capable of delivering large amounts of power.
Internal Distribution: Once electricity is supplied to the building, it must be distributed internally via robust systems of circuits, outlets, and equipment. The electrical system within the facility must handle the high load without overheating or failing.
-
Water Requirements
Water infrastructure must support significant flow rates to meet the cooling demands of AI systems. Data centers can require water flow rates as high as 500-1,000 gallons per minute (GPM), depending on size and cooling technology. Advanced facilities might reduce this demand with dry cooling or liquid immersion systems, but access to municipal water supplies or natural sources remains critical.
-
Connectivity
High-speed fiber optic infrastructure is non-negotiable for AI facilities. The ability to transmit and process data in real-time requires ultra-low latency and high-bandwidth connections. Proximity to network hubs and internet exchanges ensures efficiency and speed, reducing operational bottlenecks.
2. Building Specifications
Size: Hyperscale data centers typically require hundreds of thousands of square feet.
Power Density: Facilities must support high power densities (measured in watts per square foot).
Cooling Infrastructure: Incorporate advanced cooling systems to handle heat generated by AI operations.
Security: Facilities need robust physical and cybersecurity measures to protect sensitive data.
3. Emerging Development Corridors
Developers should target areas with growing demand for data infrastructure:
Rural Areas Near Metro Hubs: These regions offer lower land costs and easy access to power and fiber.
Decommissioned Industrial Sites: Former manufacturing or power plant sites can be repurposed for AI facilities, leveraging existing infrastructure.
Proximity to Renewable Energy Farms: Locations near wind or solar farms ensure a steady supply of sustainable power.
4. Sustainability as a Selling Point
Sustainability will be a major driver of future development. Facilities that incorporate renewable energy, water conservation technologies, and energy-efficient designs will appeal to eco-conscious investors and tenants.
Challenges and Risks
Utility Constraints
Existing utility infrastructure may not be sufficient to meet AI's growing demands. Upgrades to power grids and water systems can be costly and time-consuming.
Environmental Impact
AI facilities must balance operational efficiency with environmental concerns. Excessive water usage and carbon emissions can lead to public opposition and regulatory challenges.
Competition for Resources
As more industries adopt AI, competition for electricity, water, and fiber infrastructure will intensify, driving up costs and limiting availability.
The Road Ahead: Opportunities for Forward-Thinking Developers
AI is transforming the world—and the built environment must evolve to keep pace. Developers who understand the energy and infrastructure demands of AI have a unique opportunity to create the next generation of industrial properties.
Identify Underserved Markets: Target regions with untapped potential for utility and connectivity upgrades.
Focus on Sustainability: Incorporate green building practices to attract tenants and investors.
Leverage Partnerships: Work with utility providers, tech companies, and government agencies to develop infrastructure that supports AI growth.
The fusion of AI and real estate development represents a paradigm shift in how we think about infrastructure. By acting now, investors and developers can position themselves at the forefront of this transformative trend.
Conclusion
AI's insatiable appetite for power and water is reshaping the industrial real estate landscape. As data centers grow in size and sophistication, the demand for robust utility infrastructure will drive development in new markets. Forward-thinking developers who prioritize sustainability, utility access, and connectivity will be best positioned to capitalize on this opportunity.
The future of AI is electric—and it’s a future that smart real estate developers can help build.