Imagine you’re building a new high-tech facility—a data center, EV factory, or advanced manufacturing plant. You reach out to the utility to get electricity, only to find you’ll wait years for a grid connection and pay sky-high transmission fees once you’re in.

Now imagine someone says:

Why wait? Just put a power plant next to your building and use electricity directly.

That’s not science fiction. It’s the rapidly growing trend of behind-the-meter (BTM) co-location, and it's already reshaping how energy-intensive industries access power. These large electric users are, in effect, defecting from the grid.

What Is Grid Defection?

Grid defection refers to large customers—primarily data centers and industrial facilities—choosing to source electricity directly from a nearby generator rather than through the centralized grid. This allows them to:

• Bypass congested interconnection queues
• Avoid transmission charges (which can be 40–50% of power costs)
• Control their energy destiny with dedicated supply

In these arrangements, a customer connects to a co-located generator, such as a nuclear, gas, or solar-plus-storage plant, using protective relays to prevent drawing power from the grid. That setup keeps them technically “off-grid”—or so they argue.

The Legal Battle: Who Pays for the Grid?

As this model has grown in popularity, it’s landed squarely in front of the Federal Energy Regulatory Commission (FERC). Here’s the issue: Even if a data center doesn’t pull power from the grid, its generator often depends on the grid to stay synchronized, restart after an outage (black start), and maintain power quality (voltage and frequency). Utilities argue that co-located customers are still benefiting from shared transmission services, and should help cover the cost.

In fact, in February 2025, FERC rejected an attempt by Exelon to revise its transmission rate schedule to classify all non-islanded BTM loads as Network Load, which would guarantee they pay transmission fees. FERC ruled that PJM’s tariff lacked clarity, giving the grid operator until March 20, 2025 to either justify its current framework or propose changes (source: Wood Mackenzie).

Meanwhile, some companies argue that their customers shouldn’t pay for transmission if they’re physically isolated from the grid, even if the generator is grid-connected.

From Meta to Mainstream: The Big Tech Case for Co-Location

One of the most visible examples of this shift is Meta’s Prometheus project, a $1.25 billion AI supercomputing center in Ohio. The facility will be powered by a dedicated small modular nuclear reactor (SMR)—located adjacent to the data center but not designed to rely on traditional grid interconnection in the same way as conventional users.

Meta’s approach underscores a rising trend among hyperscale data operators: control over energy access is now as important as compute capacity. This model isn’t just about sustainability or optics—it’s about speed, resilience, and economics. With power becoming the bottleneck for AI development and cloud infrastructure, co-location is becoming a strategic imperative.

Why Customers Are Willing to Fight

For many of these companies, the math is simple:

• A 300 MW data center could pay tens of millions per year in transmission charges if connected as Network Load.
• Bypassing the grid offers faster interconnection, a critical advantage when speed to market determines competitiveness.

Even if these companies must endure multi-year legal disputes, the long-term savings are compelling enough to justify the effort.

FERC's Likely Direction: Partial Grid Charges Ahead

FERC has already suggested that co-located loads do benefit from transmission-level services, even if indirectly. These include:

1. Grid synchronization for the generator
2. Black start services in the event of outages
3. Ancillary services like voltage regulation and frequency stability

The Commission seems likely to conclude that co-located loads must contribute something toward transmission costs—just not necessarily the full freight (source: Wood Mackenzie).

In the long term, this could lead to the creation of new classes of transmission service tailored to BTM configurations.

The Real Risk Is at the State Level

While the FERC debate dominates headlines, the bigger barrier may come from state laws. In vertically integrated states or monopoly utility territories, laws often prohibit third-party generators from serving retail customers directly, even if physically adjacent.

States like Kentucky already have such laws in place, and SCE (Southern California Edison) has asked FERC to confirm that it won’t override these state protections.

Even in retail choice states, serving multiple tenants in a shared “energy park” can be tricky if utility rights-of-way are crossed. That creates challenges for large-scale co-location campuses—like those envisioned by Google, Intersect Power, and Meta.

What This Means for the Energy Transition

At Ideal Energy, we see this as a flashpoint in the evolving relationship between grid infrastructure, customer autonomy, and the future of energy markets. While grid defection isn't a solution for everyone, it raises critical questions:

• How can we accelerate grid interconnection without sacrificing system reliability?
• Should customers who only partially benefit from the grid pay full transmission rates?
• How can state and federal policies support innovation without destabilizing utility cost recovery?

Bottom Line (The Simple Version)

Big electric users are digging their own "wells" to avoid the crowded and expensive “city plumbing” of the grid. But the city (utilities) says:

“Even if you’re not drinking our water, your pipes still depend on our system—so you owe something.”

As regulators sort through the semantics, economics, and policy implications, one thing is clear: The future of grid access will not be one-size-fits-all.

Sources:

• Wood Mackenzie. (2025).
• Federal Energy Regulatory Commission (FERC)
• PJM Interconnection
• Meta's Prometheus AI Supercluster – Inc. Magazine
• Southern California Edison