Netting sounds like plumbing — a back-office accounting convenience that lets a grid operator collapse a jumble of charges and credits into one tidy number. For decades, that’s mostly what it was. But in the last three years, netting has become one of the most contested ideas in U.S. electricity regulation, showing up in fights over battery storage in Texas, hybrid solar-plus-storage projects in California, and — most consequentially — the wave of AI data centers now trying to plug directly into power plants across PJM, MISO, and the rest of the country’s organized markets.

The common thread in every one of these fights is the same question, asked in different clothes: when a facility’s own generation offsets its own load, should that offset be netted away before it ever touches the settlement or billing system — or should the two sides be metered, priced, and paid separately?

That single question turns out to hide an enormous amount of complexity, because the answer changes who pays for the grid.

Why netting matters more than it sounds like it should

Picture a facility that both generates and consumes electricity behind one meter — a battery that draws power to charge and discharge, or a data center built next to its own gas plant. There are two ways to settle its activity:

• Net it: subtract load from generation (or vice versa) and settle only the difference.
• Gross it up: meter and bill the generation and the load as two separate transactions.

If the facility is purely self-contained — its generation is only ever used to run its own equipment, and it never draws meaningfully on the wider grid — netting is a reasonable simplification. But almost nothing is that clean anymore. Batteries draw grid power to charge. Co-located generators sometimes back-feed excess power to the grid. Data centers with on-site gas turbines still lean on the transmission system for backup, black start, and regulation service even when their real-time net draw looks like zero.

Netting, in other words, can make a facility’s true use of the shared grid disappear from the accounting — even though the grid still had to be built, maintained, and kept reliable to support it. Whoever loses that accounting fight ends up paying for infrastructure someone else uses.

Example one: ERCOT and the battery that “isn’t really” drawing power

ERCOT’s fight (NPRR1194 and NPRR1197, filed 2023) was narrow but revealing. A grid-scale battery has its own “auxiliary load” — the power its cooling systems, controls, and HVAC draw just to keep the battery itself running. The battery industry and ERCOT staff argued this auxiliary load should simply be netted against the battery’s wholesale discharge, the same way a gas plant nets its own station power against its output. A transmission and distribution cooperative, South Texas Electric Cooperative, disagreed: Texas rules require a wholesale storage facility to be separately metered from its own auxiliary load, because that auxiliary load is legally a retail transaction, not a wholesale one — and retail transactions are supposed to generate retail delivery-charge revenue for utilities like STEC.

Who’s right depends entirely on which legal category you think auxiliary load belongs in. Frame it as “station power, just like any generator” and netting looks like fair, consistent treatment. Frame it as “retail load that happens to sit next to a battery” and netting looks like a subsidy, quietly moving cost off the battery owner and onto the distribution utility’s other customers. ERCOT’s stakeholder process spent over a year tabling the dispute rather than resolving it — not because the engineering was hard, but because the categorization question is really a legal and political one dressed up as a metering question.

Example two: PJM and the data center that generates its own power

PJM’s version of this fight is bigger, faster-moving, and has already produced a real rule. Under PJM’s old behind-the-meter generation (BTMG) rules, a large load co-located with its own generator could net that generation against its peak demand, cutting its transmission charges. FERC’s December 2025 order found this arrangement no longer just and reasonable, for a reason that generalizes well beyond PJM: netting large BTMG “masks” a facility’s true call on the transmission system. A 500 MW data center sitting behind a 500 MW gas plant might show up in the settlement system as a net-zero user of the grid — right up until its own generator trips offline and the full 500 MW instantly becomes the grid’s problem. FERC’s fix was a bright-line materiality threshold: below 50 MW, the old netting treatment survives; above it, the facility loses the right to net and must take one of several new, fully accounted-for transmission services, with existing arrangements grandfathered only for contracts already in place.

The stakeholder reaction shows exactly how contested the line-drawing is. Vistra and Constellation — both large generation owners who benefited from the broader co-location ruling — protested that PJM’s specific 50 MW implementation was too restrictive. The Data Center Coalition called it commercially unworkable. PJM’s own Independent Market Monitor pushed the opposite direction entirely, arguing that curtailment promises from data centers aren’t enforceable in practice and that co-located loads should simply pay for full network service, full stop. Three sophisticated parties, looking at the same 50 MW number, reached three incompatible conclusions about whether it was too strict, too permissive, or beside the point.

Example three: CAISO’s hybrid resources and the netting rule that predates the crisis

California’s version of this question is older and, in some ways, more mature — because CAISO has been dealing with hybrid solar-plus-storage resources for longer than most markets have had grid-scale batteries at all. CAISO’s tariff defines “station power” as retail energy used solely to operate a generating facility for market participation, and a long-running CPUC proceeding (R.15-03-011) has been working through exactly how that definition should apply when a single resource ID covers both a solar array and a co-located battery, each drawing and injecting at different times of day. The goal has been to give hybrid resource developers clarity before they build, rather than litigate it retroactively — which is precisely what MISO and CAISO now don’t have time to do, given FERC’s fast-moving national timeline.

Example four: the whole country, all at once

That timeline is the real story right now. On June 18, 2026, FERC issued show-cause orders to all six of its jurisdictional markets — PJM, MISO, SPP, CAISO, ISO-NE, and NYISO — demanding each one justify, within 60 days, why its tariff’s treatment of co-located load and behind-the-meter generation remains just and reasonable. This turns what had been PJM’s isolated fight into a simultaneous, market-by-market renegotiation of the same underlying question, with each region starting from a different place: SPP already has a mature large-load framework and was held up as the model; MISO admitted outright that it has no consistent framework at all, despite having the fastest data-center growth of any region; CAISO has a head start via its existing hybrid-resource proceeding but still owes FERC a full response; and ISO-NE hasn’t addressed the issue in any formal way yet.

What makes this genuinely hard, not just contentious

Strip away the jurisdiction-specific detail and the same three tensions recur everywhere this issue surfaces:

• Consistency versus specificity. Treating batteries, hybrid resources, and gas-fired data centers identically to conventional generators is administratively simple and non-discriminatory — but conventional generators don’t have anything like a data center’s scale or growth rate, so identical treatment can produce wildly different reliability and cost outcomes.
• Legal categories don’t map cleanly onto physical reality. Whether auxiliary load is “retail” or “wholesale,” and whether a materiality threshold is 10 MW or 50 MW or 500 MW, are line-drawing exercises dressed up as technical questions — and every line drawn creates winners on one side and losers on the other.
• Grandfathering is unavoidable and always contentious. Every reform proposal so far — ERCOT’s, PJM’s, and almost certainly MISO’s and CAISO’s — has had to wrestle with what to do about facilities that were financed under the old netting rules. Change the rule prospectively and you protect existing investors while leaving the underlying problem (cost shifting) in place for years. Change it retroactively and you undermine the predictability investors were relying on when they signed interconnection agreements in the first place.

None of the six pending FERC proceedings will fully resolve this by the August 17, 2026 response deadline — that date only marks when each market has to show its work, not when the fight ends. If anything, the ERCOT and PJM examples suggest the opposite: netting disputes tend to get more protracted the closer they get to a real financial number, because that’s exactly the point at which “who pays for the grid” stops being an abstraction and starts being a line item on somebody’s balance sheet.