For two years, the consensus warning on U.S. battery storage was that the grid couldn’t physically connect what developers planned to build. Interconnection queues measured in years, transformer lead times measured in weeks-into-the-hundreds, withdrawal rates above 70%. The forecast said the capacity was coming; the bottleneck data said delivery — not demand — was the binding constraint.
Now 2025 is in the books, and the verdict is more interesting than either camp expected: deployment landed almost exactly on forecast. The friction is real, but it isn’t where the headline framing put it.
The Data: What Happened
U.S. utility-scale battery storage added roughly 16 GW in 2025, with another ~3 GW behind-the-meter, for total deployment near 19 GW — about 60% annual growth and the second-largest capacity addition of any generation type after solar (Source: IEA Battery Storage commentary, June 2026, via Energy-Storage.news and SolarQuarter; EIA Preliminary Monthly Electric Generator Inventory). That outcome is the story. The EIA’s January 2025 inventory had projected 18.2–19.6 GW of utility-scale additions for the year. Actuals came in within that band. The near-term pipeline cleared.
The forward forecast keeps climbing. As of the EIA’s latest Short-Term Energy Outlook, utility-scale battery storage is projected to grow from 44,630.7 MW to 67,549.6 MW over the 12-month forecast window ending February 28, 2027 — a 51.4% increase (Source: U.S. EIA, reviewed by the SUN DAY Campaign, via Electrek, April 27, 2026, electrek.co).
The generation backdrop is shifting in tandem. Renewables’ share of total U.S. utility-scale generating capacity is projected to rise from 33.4% (March 1, 2026) to 36.6% (February 28, 2027), while natural gas’s share is forecast to slip from 40.0% to 38.3% and total fossil capacity to post a net decline of roughly 4.9 GW (Source: EIA via SUN DAY Campaign / Electrek).
Figure 1 — U.S. utility-scale battery storage, cumulative GW. Capacity reached an actual 44.6 GW in early 2026 — right on EIA’s prior forecast — with a dashed projection to 67.5 GW by February 2027. The interconnection-queue and transformer friction lands on the dashed tranche, not the solid one. Source: EIA STEO & Preliminary Monthly Electric Generator Inventory (Mar 2026); LBNL Queued Up 2025; Wood Mackenzie / POWER.
Why 2025 Delivered: A Population Problem
The reason the “delivery is the constraint” thesis didn’t bite in 2025 comes down to a distinction that’s easy to miss: the EIA’s near-term planned additions and the much-cited interconnection queue are not the same set of projects.
The EIA’s planned-additions data comes from the Preliminary Monthly Electric Generator Inventory (Form EIA-860M) — projects that have reported firm in-service dates. These are late-stage projects: signed interconnection agreements, secured equipment orders, financing largely in place. They have already survived the gauntlet.
The interconnection queue is the opposite end of the funnel. LBNL data shows roughly 890 GW of storage capacity actively seeking interconnection as of end-2024, against a historical reality where more than 70% of all requests are ultimately withdrawn (Source: LBNL, Queued Up 2025 Edition, December 2025, emp.lbl.gov). That 70% attrition is a fact about the speculative pile — projects that may never break ground — not about the advanced projects the forecast actually counts.
Applying queue-wide attrition to near-term, firm-date additions conflates the two. The 2025 result is the evidence: the projects EIA counted got built, because they were the ones that had already cleared the slow part.
Where the Friction Actually Bites: The 2027+ Tranche
This doesn’t mean the bottleneck story is wrong. It means it’s mis-located in time. The delivery friction is real and structural — it just constrains the projects entering the queue now, which feed the 2027 and later vintages, not the ones already in the ground.
Value chain anatomy. A utility-scale storage installation breaks into distinct segments, each with its own supply chain and lead times: battery cells and packs (lithium-ion dominant at ~99% per LBNL), inverters and power conversion, balance-of-system (thermal management, racking, switchgear, distribution), and grid interconnection hardware (transformers, switchgear, transmission upgrades). The interconnection hardware segment is where the timeline stretches.
The queue clock. The median time from interconnection request to commercial operation has more than doubled — from 22 months for projects built in 2008 to 55 months (about four and a half years) for projects completed in 2024 (Source: LBNL, Queued Up 2025 Edition). A project filing today is, on the historical median, a 2029–2030 in-service date. That math is what governs the tranches beyond the current forecast window, not the one within it.
Component lead times. Power transformers averaged 128-week lead times and generator step-up transformers 144 weeks as of Q2 2025, with high-voltage switchgear around 44 weeks — versus a 7–14 month pre-pandemic norm for large transformers. Roughly $2 billion has been committed to new or expanded North American transformer manufacturing since 2023, but the major new plants don’t come online until 2027–2028 (Hitachi Energy Virginia in 2028; Eaton South Carolina in 2027; Siemens Energy North Carolina in early 2027) (Source: Wood Mackenzie via POWER Magazine, “Transformers in 2026,” powermag.com).
Map those timelines against the queue clock and the picture resolves: the equipment relief arrives in 2027–2028, roughly when the projects now entering the queue would need it. The system isn’t failing to deliver — it’s delivering on a lag, and the constraint moves down the pipeline rather than disappearing.
Risks and Counterpoints
The forward forecast still assumes the advanced-stage pipeline behaves the way 2025’s did. The main risk is that the near-term population thins faster than expected — a wave of late-stage cancellations driven by escalating grid-upgrade cost allocations, policy shifts on storage incentives, or battery input cost spikes would pull actuals below the 67.5 GW forecast in a way 2025 didn’t preview. A counter-risk runs the other way: if transformer capacity comes online ahead of schedule and queue reforms (cluster studies, withdrawal penalties) accelerate, the later tranches could compress their timelines and the 2027+ friction could prove shorter-lived than the 55-month median implies. The honest read is that 2025 validated the near-term forecast and said very little about 2028 — that’s where the open question now sits.
This is an AI Briefing — AI-generated analysis published under TLCapital.AI. It is not personal research or positions, and it is not investment advice. Figures are sourced to primary filings and reporting with dates noted throughout. Do your own diligence.