In 2023, several U.S. utilities began reporting a surge in large-load interconnection requests driven by AI data centers. Some regional grid operators have publicly stated that single-site load requests now reach into the hundreds of megawatts. The International Energy Agency projects global data center electricity demand could more than double by 2026, largely driven by AI workloads.
That growth does not begin inside the data hall.
It begins at the utility.
And this is where many AI projects quietly encounter their first real constraint.
The Interconnection Reality Most Teams Underestimate
Picture a 120 MW AI deployment entering the planning phase.
Preliminary modeling assumes a service entrance built around historical fault levels. Engineering specifies UL-listed service gear and downstream UL891 switchboards based on early utility data.
Months later, the utility confirms the final interconnection design:
- Larger primary transformer
- Reinforced feeder
- Increased available fault current at the service entrance
The facility load requirement did not change.
The electrical boundary conditions did.
That change cascades.
Available short circuit current rises. Breaker interrupting ratings must be revalidated. Bus bracing requirements increase. Arc flash calculations shift. PPE classifications may change. Service entrance equipment may need to be upgraded.
What looked compliant during schematic design now requires reevaluation.
This is not unusual. It is becoming common.
Why AI Loads Change Utility Behavior
AI facilities compress significant demand into a single site. Compared to traditional enterprise data centers, AI deployments often exhibit:
- Higher sustained utilization
- Faster load ramp-up
- Less load diversity
- Tighter uptime requirements
Utilities respond by strengthening upstream infrastructure.
Stronger infrastructure often means higher available fault current at the secondary.
Here is how that plays out:
|
Design Variable |
Traditional Enterprise DC |
High-Density AI Facility |
|
Typical Rack Density |
~7–10 kW |
20–100+ kW |
|
Transformer Sizing |
Moderate |
Large, growth-oriented |
|
Available Fault Current |
Lower baseline |
Elevated due to transformer capacity |
|
Arc Flash Energy |
Manageable |
Potentially higher at service entrance |
|
Expansion Frequency |
Moderate |
Expected and planned |
When transformer size increases, fault contribution increases. That impacts every downstream assembly.
UL891 switchboards are evaluated for specific short circuit current ratings and mechanical withstand capability. Those ratings must exceed available fault current at installation.
If the utility’s final design shifts that number upward, the switchboard rating must follow.
Service Entrance Design Is Now Strategic
Transformer sizing is no longer just about meeting present load. It influences:
- Secondary fault current
- Breaker frame size
- Coordination strategy
- Arc flash incident energy
- Long-term expansion flexibility
Oversizing for growth increases fault exposure. Undersizing constrains expansion.
At higher available fault levels:
- Mechanical forces on bus bars increase during fault events
- Breaker interrupting ratings must increase
- Arc flash boundary distances may expand
- Maintenance procedures become more complex
NEC 110.9 and 110.10 require equipment interrupting and withstand ratings to exceed available fault current. Compliance is binary. Either the assembly is rated for the condition or it is not.
In high-density AI environments, fault levels can exceed early conceptual modeling. If that shift occurs after equipment procurement, redesign becomes expensive.
Coordination and Arc Flash: Where Small Errors Compound
Higher fault current does not only affect service gear.
It influences:
- Selective coordination curves
- Downstream breaker performance
- Arc flash incident energy
- Clearing times
If coordination studies are not updated after final utility data is confirmed, protection strategy can drift out of alignment.
Arc flash incident energy is particularly sensitive to fault current magnitude and clearing time. Elevated fault levels increase potential energy release.
From an operational standpoint, this affects:
- Worker safety requirements
- PPE classification
- Maintenance intervals
- Insurance assessments
A service entrance that looks straightforward at 35 kA behaves differently at 65 kA.
Why Early Engineering Changes Project Viability
Interconnection constraints are increasingly defining project timelines.
Some utilities now require extended review for large-load applications. Substation upgrades and feeder reinforcements can add months to deployment schedules.
Electrical design that anticipates utility evolution avoids reactive redesign.
When manufacturing partners are brought in early, several risks can be mitigated:
- Short circuit rating alignment before fabrication
- Realistic bus bracing selection
- Breaker interrupting margin validation
- Coordination studies built on confirmed fault levels
- Expansion headroom designed intentionally
This shifts the manufacturer from downstream installer to upstream engineering partner.
In AI infrastructure, that shift matters.
The Strategic Takeaway
AI is concentrating electrical demand in ways the grid is still adapting to.
Utility interconnection is no longer a procedural step. It is a design constraint.
Transformer sizing, fault current availability, service entrance configuration, and coordination strategy determine:
- Whether inspection passes cleanly
- Whether arc flash exposure is manageable
- Whether expansion is feasible without replacement
- Whether the schedule holds
Projects that align utility studies, NEC requirements, and UL assembly ratings early operate with more control.
Projects that assume those variables remain static discover otherwise.
If you are evaluating a large-load interconnection, reviewing transformer strategy, or validating short circuit exposure before releasing equipment, upstream engineering makes a measurable difference.
To connect with Moonshot’s electrical engineering team and review your interconnection approach:
https://moonshotus.com/request-form/
The grid sets the boundary conditions. Your design determines how well you operate within them.