How a Battery Became a Data Center's Fastest Way Onto the Grid
How a Battery Became a Data Center's Fastest Way Onto the Grid
A developer I spoke with earlier this year put it in a way that stuck with me: the battery on his data center project wasn't there in case the grid failed. It was there so the grid would say yes.
That's a strange thing to hear if you've spent any time around this equipment, because storage has always been sold as insurance. Something you install so the servers stay up when the utility has a bad day. But talk to anyone actually trying to get a large load connected in the US right now, and you'll hear a different story. The battery isn't the backup plan anymore. Sometimes it's the whole plan.
What's actually happening at the interconnection desk
US utilities are on track to add roughly 24 gigawatts of new battery storage in 2026, and a lot of that isn't demand growth in the usual sense — it's utilities responding to how badly the interconnection queue has backed up. PJM, which runs the grid covering the heaviest data center corridor in the country, posted a 6,623 megawatt capacity deficit in its last auction, and the clearing price hit a record. Since 2020, roughly three out of four projects studied in PJM's queue have simply withdrawn before ever getting built.
Texas tells a slightly different version of the same story. ERCOT doesn't have the queue backlog PJM has, but it has its own scarcity problem, and some of the largest standalone battery projects heading toward commercial operation in 2026 and 2027 sit near the Dallas and Houston data center clusters specifically because that proximity works two ways — merchant revenue when the grid needs the power, and a direct offtake arrangement when a data center next door does.
Washington noticed. In October 2025 the Department of Energy asked FERC to take jurisdiction over how large loads connect to the bulk transmission grid and to standardize the process nationally, with a target date of April 30, 2026 for FERC's final action on it. However that plays out, the fact that it needed to happen at all tells you the old case-by-case interconnection process wasn't built for what's showing up at the front door now.
Bring your own capacity, explained plainly
The term making the rounds is "bring your own capacity," or BYOC. The idea is simple enough even if the engineering behind it isn't: a developer pairs their load with enough on-site battery storage that they can commit to drawing less from the grid during the hours it's most stressed, and in exchange, the utility can approve a faster or larger interconnection than it otherwise would. Pair that with what people are calling flexible grid connections, and batteries end up doing something gas turbines structurally can't — they can absorb and release power on a timescale that matches how congestion actually shows up, which according to the Department of Energy's own transmission needs research is concentrated in a small share of hours, not spread evenly across the year.
None of this is a workaround anymore. NEMA published an actual standard for it in January — US 80074-2025, covering how to design battery storage into a data center's power topology, along with a companion guide on microgrid integration. When a standards body writes the reference architecture, that's usually the signal an approach has stopped being experimental.
What it means for how a BESS gets specified
If a battery's job now includes getting a project connected, not just keeping it running afterward, the sizing conversation changes. It's no longer just "how many hours of backup does the load need." It becomes a question of discharge and recharge profile against a specific node's congestion pattern, how the system talks to the utility's interconnection agreement, and whether the equipment and its documentation can support whatever verification the utility ends up asking for. That's a harder spec to write, and a harder one to build storage against.
We build our CDEU commercial and industrial storage systems, ranging from 71 to 160kWh, with exactly that kind of project in mind — the kind where the battery is doing real grid work, not sitting in a room as a contingency. It's a different design brief than storage that only has to work when everything else has already failed.
A few questions we keep getting asked
Is BYOC only relevant in PJM and ERCOT? Those two happen to have the clearest public data, but developers in other constrained markets are asking utilities the same question, because the underlying problem — a grid that can't approve new large loads fast enough — isn't unique to any one region.
Does a BYOC battery need to be sized differently than a standard backup system? Generally yes. A backup system is sized around how long it needs to cover a load during an outage. A BYOC system is sized around how much load it can shift off the grid during the hours the interconnection agreement actually cares about, which is a narrower and more specific question.
How new is this as a strategy? New enough that it was mostly informal until this year. NEMA's January 2026 standard (US 80074-2025) is the clearest sign it's moved from a handful of developers improvising to something the industry is building shared rules around.
Sourcing storage for a US data center project that needs to move fast on interconnection? [Talk to our team →]
