Co-located data center power refers to the practice of placing data centers directly at or adjacent to power generation facilities, allowing them to consume electricity behind-the-meter without relying on traditional grid transmission. This approach bypasses grid interconnection delays while providing reliable, cost-effective power for energy-intensive AI and cloud computing operations.
Key Takeaways
- U.S. electricity demand is projected to rise 2.4% by 2030, with nearly half driven by data centers requiring over 50 GW of new capacity
- Co-location enables data centers to access power directly from generation sources, avoiding grid bottlenecks and interconnection delays
- FERC is developing new rulemaking for co-located loads over 20 MW, with final rules due April 2026
The Grid Capacity Crisis Driving Co-Location
After decades of flat electricity demand, the United States faces an unprecedented surge in power consumption. According to industry projections, U.S. electricity demand will rise 2.4% by 2030, with data centers accounting for nearly half of this increase. These facilities will require over 50 GW of new capacity—more than Pennsylvania’s current total electricity output.
This explosive growth stems primarily from artificial intelligence workloads and cloud computing expansion. Traditional data centers already consume substantial power, but AI training and inference operations require exponentially more electricity. A single large language model training run can consume as much power as thousands of homes over several months.
The challenge extends beyond raw capacity. Grid interconnection queues have become severely backlogged, with new generation projects waiting years for approval and connection. Transmission infrastructure, built for a different era of electricity demand patterns, cannot accommodate the rapid deployment timelines that hyperscale data center operators require.
Why it matters for builders: Co-location transforms stranded or underutilized generation assets into high-value infrastructure, creating new revenue streams without grid expansion costs.
How Co-Location Works in Practice
Co-located data center power arrangements place computing facilities directly at generation sites, consuming electricity before it enters the transmission grid. This behind-the-meter approach eliminates transmission losses, reduces grid congestion, and provides data centers with direct access to reliable baseload power.
The most prominent example involves Microsoft’s partnership with Constellation Energy to restart the Three Mile Island nuclear plant. This 835 MW facility will provide dedicated power to Microsoft’s data centers, demonstrating how co-location can revive dormant generation assets while meeting clean energy commitments.
Major technology companies with a combined market capitalization of $7 trillion—including Microsoft, Google, and Amazon—are actively pursuing co-location strategies. These arrangements span multiple generation types: nuclear plants for baseload reliability, hydroelectric facilities for clean power, natural gas plants for flexible capacity, and renewable installations paired with storage systems.
Economic and Operational Benefits
Co-location delivers compelling economic advantages for both data center operators and power generators. Data centers gain access to wholesale electricity rates, avoiding retail markups and transmission charges. They also secure long-term power purchase agreements that provide cost certainty for capital-intensive computing infrastructure.
For power plant operators, co-location creates new revenue streams that can justify keeping existing facilities online or financing new construction. Plants with surplus capacity can monetize excess generation without competing in volatile wholesale markets. This arrangement proves particularly valuable for nuclear facilities, which face economic pressure despite their clean energy credentials.
| Generation Type | Co-Location Benefits | Typical Capacity | Key Considerations |
|---|---|---|---|
| Nuclear | 24/7 baseload, zero emissions | 800-1,200 MW | Regulatory complexity, high capital costs |
| Natural Gas | Flexible dispatch, rapid deployment | 200-800 MW | Fuel supply, emissions considerations |
| Hydroelectric | Clean energy, long asset life | 50-500 MW | Environmental permitting, seasonal variation |
| Renewables + Storage | Clean energy, declining costs | 100-400 MW | Intermittency, storage duration limits |
Regulatory Framework and FERC Response
The Federal Energy Regulatory Commission is developing comprehensive rulemaking to address co-located loads over 20 MW. Directed by the Department of Energy in October 2025, these rules must be finalized by April 2026. The regulatory framework will establish standards for interconnection agreements, cost allocation, and grid reliability impacts.
According to the Electric Power Supply Association, current regulatory uncertainty creates project development risks. FERC scrutiny of interconnection agreements could delay projects or impose strict modifications, particularly affecting hydroelectric facilities where environmental and operational constraints add complexity.
The rulemaking process must balance competing interests. Utilities worry about resource adequacy if existing generation becomes dedicated to specific loads rather than serving the broader grid. Data center developers seek regulatory clarity to support long-term investment decisions. Power generators want flexibility to optimize revenue streams across wholesale markets and direct sales.
Grid Impact and Resource Adequacy Concerns
Critics of co-location raise legitimate concerns about grid reliability and resource adequacy. When existing power plants dedicate capacity to co-located data centers, that generation may no longer be available to serve broader grid needs during peak demand periods or emergency conditions.
However, proponents argue that co-location often involves surplus generation capacity that would otherwise remain unused. Plants operating below full capacity can serve co-located loads without reducing grid supply. Additionally, many co-location projects involve new generation construction specifically designed to serve data center loads while maintaining grid contributions.
The efficiency argument carries significant weight. Co-located arrangements eliminate transmission losses, which typically range from 5-8% of generated electricity. For large data centers consuming hundreds of megawatts, these efficiency gains represent substantial energy savings and reduced environmental impact.
Tools & Resources
- Energy market data & stock screening — Track publicly traded utilities and power generators pursuing co-location strategies.
- Financial news & market analysis — Monitor regulatory developments and industry partnerships affecting co-location projects.
Future Outlook and Market Evolution
Co-location represents a fundamental shift in how electricity infrastructure serves digital economy demands. As AI workloads continue expanding and grid constraints persist, behind-the-meter arrangements will likely become standard practice for hyperscale data center development.
The model’s success depends on regulatory clarity, technological advancement, and continued cost competitiveness. Nuclear co-location projects face the longest development timelines but offer the most reliable baseload power. Natural gas facilities provide faster deployment but must navigate emissions regulations. Renewable co-location requires storage integration to match data center uptime requirements.
Market dynamics favor continued co-location growth. Technology companies possess the capital and long-term planning horizons necessary for direct power procurement. Power generators seek stable revenue streams to support infrastructure investments. Grid operators benefit from reduced transmission congestion and improved system efficiency.
FAQ
What is co-located data center power?
Co-located data center power involves placing data centers directly at power generation facilities to consume electricity behind-the-meter, bypassing traditional grid transmission and avoiding interconnection delays.
Why are companies choosing co-location over traditional grid connections?
Co-location eliminates grid interconnection delays, provides more reliable power supply, reduces transmission costs, and enables faster deployment of energy-intensive AI and cloud computing infrastructure.
How does FERC regulate co-located data centers?
FERC is developing new rulemaking for co-located loads over 20 MW, with final rules due April 2026. The regulations will address interconnection standards, cost allocation, and grid reliability impacts.
What types of power plants work best for data center co-location?
Nuclear plants offer 24/7 baseload power, natural gas provides flexible dispatch, hydroelectric delivers clean energy, and renewables paired with storage can meet sustainability goals while maintaining reliability.
Co-located data center power represents a pragmatic response to unprecedented electricity demand growth and grid infrastructure constraints. While regulatory frameworks continue evolving, the fundamental economics and operational benefits make co-location an increasingly attractive solution for both technology companies and power generators. Success will depend on balancing grid reliability concerns with the need for efficient, scalable electricity infrastructure that can support the digital economy’s continued expansion.