Key takeaways
- Two feeds are only redundant if they do not share a substation, structure, route or single point of failure.1
- Common configurations range from a radial tap or a ring-bus loop-in of one line to two lines from separate substations.3
- On-site redundancy (N+1, 2N, generators, UPS) and utility source redundancy solve different problems; Uptime Institute’s Tier approach treats on-site generation, not the utility, as the reliable source.45
- Two sources on the same network can see the same disturbance: in July 2024 about 1,500 MW of Northern Virginia data center load disconnected after repeated faults on one 230 kV line.2
- A second source often needs its own easements, substation bay and land on the campus.
- Many campuses start on one source and add the second for a later phase.
01Why data centers want two utility sources
A data center is built to keep running through equipment failures. Inside the fence that means UPS systems, backup generators and redundant electrical paths. Uptime Institute’s Tier requirements go further: they describe on-site generation as the only source of reliable power, and for Tier III and IV designs the generator, not the utility, is treated as the primary source, sized to carry the full load without a limit on run hours.45 In practice, most data center generators are standby units that run little when a utility is present.5
Running on generators is still something operators work hard to avoid. Under federal engine rules, emergency generators are generally limited to about 100 hours a year of maintenance, testing and other non-emergency use, while operation during a true emergency is not capped.67 Long outages also mean fuel logistics and, depending on the permit, local emissions limits. Our guides to backup generators and fuel sizing and air permits for generators cover those limits.
Two independent utility sources reduce how often the campus has to lean on its generators. If one line, transformer or substation is out for a fault or for maintenance, the other carries the campus. For large campuses, a second source also gives the utility flexibility to maintain its own equipment without asking the customer to go to generators.
Not every project needs it. Some operators accept a single source and rely on on-site systems. Others treat a second source as a hard requirement for any site above a certain size. The requirement should come from the operator’s design standard, because it changes the substation, transmission and land plan.
02What makes a second feed truly independent
Two lines arriving at a site are not automatically two sources. The test is whether any single event could take out both. A common high-availability design specifies two primary utility feeders from separate substations that follow independent routes to the site, each sized to carry the entire load on its own.1 Utilities and engineers look for common points of failure at every level.
- Different source substations, ideally fed from different parts of the transmission network.
- Separate circuits, not two conductors on one set of towers or poles.
- Separate routes (route diversity), so one storm path, fire, flood or vehicle strike cannot reach both lines.
- Separate entry points to the campus and separate positions in the customer substation, continuing into separate medium-voltage switchgear lineups.1
- No shared upstream element, such as one transformer or one line, that both feeds depend on.
Fig. 1Where redundancy has to hold, grid to IT load
- 01Transmission networkIdeally two parts of the grid, not one corridor
- 02Source substations A and BSeparate substations, not one with two breakers
- 03Lines on separate routesSeparate structures, easements and entry points
- 04Customer substationTwo line positions, transformers sized for the load
- 05MV switchgear A and BPaths stay separate toward UPS and generators
- 06IT loadServed through on-site N+1 or 2N systems
Perfect independence is rare. Most real designs accept some shared exposure and document it. The point is to know where it is. A loop-in of one line gives two paths into the site but still depends on that line’s corridor and the substations at each end. Even two separate substations sit on one regional grid: one Texas colocation provider describes its building as served by two feeds from two independent substations, each connected to the ERCOT grid, which protects against a local failure but not a grid-wide event.8
03Common source configurations
The configuration available at a site depends on what lines and substations are nearby and what the utility will accept. These are the patterns that come up most.
| Configuration | What it is | Resilience | Typical cost and land |
|---|---|---|---|
| Single radial tap | One line extended from an existing line | Lowest; one path | Lowest |
| Loop-in of one line | Existing line cut into a new switching station, often a three-breaker ring bus, fed from both ends39 | Better; two directions, shared corridor | Moderate; switching station footprint |
| Two circuits, one corridor | Two separate circuits on separate structures sharing a right-of-way | Good, with shared route exposure | Moderate to high |
| Two lines from separate substations | Independent lines from two different source substations | High | High; two line routes and easements |
| Two voltage levels or networks | Sources from different parts of the grid, such as two transmission systems | Highest, where available | Highest; complex studies |
Utility filings show what a loop-in looks like in practice. PJM facilities studies for generator interconnections in Dominion Energy’s territory, which are public, describe new 230 kV three-breaker ring bus switching stations that cut into an existing line, dividing it into two segments that each terminate at the new station.39 In Loudoun County, Dominion proposed a 230 kV double-circuit line and the Evergreen Mills switching station to serve data centers in Arcola, with the transmission loop built in two phases.10 Its Aviator project paired a new substation south of Dulles Airport with transmission lines to serve both the substation and a new data center, approved by Virginia’s State Corporation Commission in November 2022.11
For the basics of how a campus connects to a line, see our guide to transmission voltage for data centers. The substation side is covered in substation capacity and proximity and substation configurations for data center campuses.
04Utility redundancy vs. on-site redundancy
These are often confused. On-site redundancy, described as N+1 or 2N, is about the transformers, switchgear, UPS and generators inside the campus. It protects against failure of the customer’s own equipment and carries load through outages. Utility source redundancy is about how many independent grid paths reach the campus. Our guide to data center tier levels and redundancy covers the on-site side.
Fig. 2Utility source vs. on-site redundancy
Utility source redundancy
- Two or more independent grid paths
- Covers loss of a line, transformer or substation
- Needs easements, line routes and studies
- Does not cover a regional grid event
N+1, 2N
On-site redundancy
- Redundant transformers, switchgear and UPS
- Generators carry load through outages
- Can be built on a single utility source
- Limited by fuel, permits and maintenance
- A 2N electrical design does not double grid demand and does not require two utility sources by itself.
- Two utility sources do not replace generators or UPS; a regional event or a voltage disturbance on the shared network can affect both sources.
- A dual-source campus may be able to size or run its generators differently, but that is a design choice, not a given.
- Utilities size each source to carry the full campus load or a defined share of it, which affects the cost of the second source.
The 2024 Virginia event shows the limits. A failed lightning arrester on a 230 kV line led to six successive faults in 82 seconds as automatic reclosing tried to restore the line; the faults were cleared normally, but customer-side protection at data centers moved about 1,500 MW of load off the grid.2 The load loss spanned roughly 60 load points across 25 substations, so many sites with their own sources still saw the same voltage disturbance.1213 PJM later reported a similar transfer of about 1,500 MW on February 17, 2025, and a larger event of nearly 4,000 MW on July 22, 2025, the largest in its history, and is evaluating ride-through standards for large loads.14 Our guide to NERC reliability and large loads follows that work.
Fig. 3The July 2024 Northern Virginia load loss
- data center load that left the grid
- ~1,500 MW
- in an 82-second period on one 230 kV line
- 6 faults
- load points across 25 substations
- ~60
Clarify early whether the second source must carry the whole campus alone or only a share. A source sized for full load costs more and may need larger upgrades than one meant to carry critical load only.
05What a second source costs in land, money and time
The second source is often the more expensive and slower one, because the first source usually takes the easiest path. The second may need a longer line, a different substation with less room, or network upgrades on a different part of the system.
- Easements. A second line on a separate route usually crosses different landowners, which adds negotiation and possibly condemnation by the utility where state law allows it. See eminent domain and transmission easements.
- Campus land. A second line entry, a larger customer substation or a separate switching station takes acreage and affects building layout.
- Studies. Each source is studied for its own capacity and contingency limits. The second may trigger upgrades the first did not.
- Cost responsibility. Whether the customer pays directly for the second source or it is treated as a network improvement depends on utility rules. In one PJM generator interconnection study in Dominion’s territory, the change-of-ownership point sits at the shared fence line, with the utility owning and maintaining its side.9 See transmission upgrades and who pays for them.
- Schedule. Long-lead equipment applies to both sources. Wood Mackenzie’s 2025 survey put average power transformer lead times at about 128 weeks, and a separate line route may face its own siting review.15
06How to screen a site for redundant feeds
- 01Confirm whether the operator’s design standard requires two sources, and whether each must carry full load.
- 02Map every transmission line and substation within a practical distance, and note which substations feed which lines.
- 03Identify at least two paths that do not share a substation, structure or corridor.
- 04Estimate the line route and easement count for each path.
- 05Reserve land on the campus for two line entries and the substation or switchyard layout they imply.
- 06Ask the utility whether a phased approach, one source first and the second later, is acceptable, and what ride-through and protection settings it expects from the campus.
Phasing the second source is common. A first building may run on a single source while the second line is built, provided the operator accepts that interim risk. The utility’s answer to that question can move a site’s first power date by years, which is why we treat it as part of the load ramp schedule. Who builds and owns the customer substation also changes the plan; see utility-owned vs. customer-owned substations. To check a specific parcel, you can get a site reviewed.
Common questions
Do data centers need two utility feeds?
Many large data centers want two independent utility sources, but it is a design choice, not a universal rule. Uptime Institute’s Tier approach treats on-site generation, not the utility, as the primary source of reliable power, so the number of utility feeds is not what earns a Tier rating.45 Larger campuses and operators with strict uptime standards commonly require two sources, and sometimes require each to carry the full campus load.
What is a dual substation feed?
A dual substation feed means the campus is served by lines from two different source substations rather than one. It protects against the loss of a single substation, transformer or line. The two feeds are most independent when they come from separate parts of the transmission network, run on separate structures and routes, and enter the campus at separate points.1
Is a loop-in the same as two independent sources?
Not quite. A loop-in cuts an existing transmission line into a new switching station, often a ring bus, so the site is fed from both directions.3 That protects against a fault on one section of the line, but both paths still share the same corridor and depend on the same line and its end substations. It is more resilient than a radial tap and less independent than lines from two separate substations.
Does 2N redundancy require two utility sources?
No. 2N describes the on-site electrical design, with two complete power paths from the customer substation to the IT equipment. It can be built on a single utility source. Two utility sources address a different risk: loss of the grid connection itself. Many operators want both, but they are separate decisions with separate costs.
Do two utility feeds protect against grid voltage disturbances?
Not reliably. Both feeds are usually on the same regional network, so a nearby fault can depress voltage on both. In the July 2024 Northern Virginia event, about 1,500 MW of data center load transferred off the grid after repeated faults on a single 230 kV line, even though the faults were cleared normally.213 Ride-through settings matter as much as the number of feeds.
Notes
- 1.EC&M, “A High Availability Electrical Design for Today’s Data Center,” n.d. ecmweb.com
- 2.North American Electric Reliability Corporation, “Incident Review: Considering Simultaneous Voltage-Sensitive Load Reductions,” 2025. nerc.com
- 3.PJM Interconnection, “AE1-153 Facilities Study,” n.d. pjm.com
- 4.Uptime Institute, “Tier Requirements for Power,” n.d. uptimeinstitute.com
- 5.Consulting-Specifying Engineer, “Generator ratings and the implications for data centers,” n.d. csemag.com
- 6.Consulting-Specifying Engineer, “Understanding emission requirements for standby gensets,” n.d. csemag.com
- 7.U.S. Environmental Protection Agency, “2012 Amendments to the Stationary Engine NESHAP and NSPS,” n.d. epa.gov
- 8.Data Foundry, “Dual Power Feeds,” 2024. datafoundry.com
- 9.PJM Interconnection, “AG1-098 Facilities Study, Transition Cycle 1, Phase 3,” n.d. pjm.com
- 10.Data Center Dynamics, “Dominion proposes 230kV transmission infrastructure for Arcola data centers,” n.d. datacenterdynamics.com
- 11.Dominion Energy, “Aviator,” n.d. dominionenergy.com
- 12.North American Electric Reliability Corporation (presentation hosted by FERC), “NERC Seeks to Address Reliability Impacts from Large Load Integration,” 2025. ferc.gov
- 13.Data Center Dynamics, “Virginia narrowly avoided power cuts when 60 data centers dropped off the grid at once,” 2025. datacenterdynamics.com
- 14.American Public Power Association, “PJM, Dominion review large load transfer event,” 2025. publicpower.org
- 15.pv magazine USA, “U.S. transformer market faces severe supply constraints as lead times extend to four years,” 2026. pv-magazine-usa.com
Have a site in mind?
Get a straight answer on your land.
Send a parcel number, an address, a map pin or a target load. We’ll tell you what it can support and what it would take.
This guide is general information about data center site selection. It is not engineering, legal, tax or investment advice. Requirements vary by state, utility and county, so confirm the specifics for any site with the relevant authorities and advisors.
Terms in this guide
Related guides
More in Power & interconnection
- How Much Power Does a Data Center Need?
- Power Timelines: Why “When” Matters as Much as “Where”
- ISOs, RTOs and Utility Territories in Data Center Siting
- On-Site Generation and Bridge Power for Data Centers
- What Is Powered Land?
- Load Ramp Schedules: How to Phase Power for a Data Center Campus
- Large-Load Tariffs and Electric Service Agreements, Explained
- Behind-the-Meter vs. Front-of-the-Meter Power for Data Centers
- Microgrids and Battery Storage at Data Center Sites
- Co-Locating Data Centers With Power Plants
- Nuclear and Small Modular Reactors: Siting Considerations for Data Centers
- Renewable Energy Procurement and How It Affects Data Center Siting
- What Is a Utility Load Study for a Data Center?
- Firm vs. Interruptible Electric Service for Data Centers
- Flexible Data Center Loads and Demand Response
- Texas Senate Bill 6: What It Means for Data Centers and Large Loads
- PJM Capacity Prices and Data Center Load Growth
- Contribution in Aid of Construction (CIAC) and Line Extensions for Data Centers
- Transformer and Switchgear Lead Times for Data Centers
- Electricity Rates and Power Costs for Data Center Sites
- Power Purchase Agreements for Data Centers: Physical, Virtual and Sleeved
- 24/7 Carbon-Free Energy and Hourly Matching for Data Centers
- Gas Turbines vs. Reciprocating Engines for Data Center On-Site Power
- Fuel Cells for Data Center Power: How They Work and Where They Fit
- NERC Reliability Standards and Large Data Center Loads: Ride-Through, Load Loss and Registration
- From Signed Agreement to Energization: The Data Center Power Timeline
- Hydropower and Data Center Siting: Public Power, Allocations and Drought Risk
- Geothermal Power for Data Centers: Enhanced Geothermal, Clean Tariffs and Siting in the West
- Surplus Interconnection Service and Retired Power Plant Sites: How Existing Grid Rights Get Reused
Guide 10 of 35 in Power & interconnection · all 242 guides · glossary · calculators
