US Battery Storage Capacity, Cost and Interconnection Queue by State

Executive Summary

EIA's Electric Power Monthly Table 6.2.A records 53.38 GW of U.S. utility-scale storage in July 2026 under its “Other Energy Storage” category, against approximately 35.51 GW in July 2025. The resulting 50.3% year-on-year change is a calculated comparison of two EIA snapshots, not a forecast. Because EIA reports hydroelectric pumped storage separately, this analysis never uses pumped-hydro capacity as a proxy for batteries.

Storage is highly concentrated. Texas has 18.02 GW and California 16.18 GW; together they hold about 64% of the installed total. Arizona adds 6.65 GW. The concentration reflects renewable buildout, system needs, market structures, procurement and project timing, not a single national driver.

The development pipeline is much larger than the operating fleet. NextMSC's state dataset records 391.50 GW of standalone battery queue capacity and 431.55 GW of solar-plus-battery primary-component capacity, or 823.05 GW in its combined battery-related view. Berkeley Lab separately reports approximately 749 GW of storage in active transmission interconnection queues at the end of 2025. These are different accounting constructs and must not be substituted for one another.

The cost layer is intentionally modeled. The U.S. base is $308/kWh for a 4-hour utility-scale system, equal to the 2026 mid-case in NREL's 2025 cost projections. State proxies scale that base by a Macro Cost Pressure Index. The results are comparison indicators, not observed turnkey prices.

Our reading: installed capacity, queue capacity and cost are three different lenses. Installed capacity describes what is operating, the queue describes requests for interconnection, and the cost proxy describes relative state cost pressure. Combining them without definition would overstate the certainty of the pipeline.

Sources: U.S. EIA (Electric Power Monthly Table 6.2.A), Lawrence Berkeley National Laboratory (Queued Up: 2026 Edition), GridAlmanac state presentations derived from LBNL, NREL (2025 cost update) and NextMSC analysis.

U.S. Battery Storage at a Glance

The July 2026 EIA snapshot is the installed-capacity anchor. Table 6.2.A reports net summer capacity by technology and state and identifies pumped storage and “Other Energy Storage” separately. NextMSC uses the latter, a deliberately conservative choice that stops the analysis from converting all storage technologies into batteries.

The change from about 35.51 GW to 53.38 GW is large in percentage terms because the starting base was materially smaller than today's fleet. State results are even more sensitive to small bases, as the next section shows.

MW, MWh and duration

Battery storage has two dimensions that should never be collapsed into one number. MW is the maximum power a system can deliver at a point in time, and MWh is the energy it can deliver at that power over its duration. A 100 MW battery with four hours of duration has 400 MWh of energy capacity, and a 100 MW battery with one hour has 100 MWh.

The distinction matters here because interconnection queues are reported in MW, since grid studies focus on power injection, while cost studies often use $/kWh because energy capacity is a major cost driver. A 4-hour $/kWh benchmark cannot be multiplied by MW alone without specifying duration. NREL's framework separates power-related and energy-related cost components, and this article uses a 4-hour system only as a common comparison point, without implying that every queued project is a 4-hour system.

Installed Storage Capacity by State

Texas and California form the centre of the installed fleet. Their combined 34.20 GW is about 64% of the U.S. total. Arizona is third at 6.65 GW. Nevada (2.09 GW), New Mexico (1.36 GW) and Florida (1.20 GW) form a second cluster, and most other states hold less than 1 GW. A battery can earn value from energy arbitrage, capacity, ancillary services, renewable integration and local reliability, and the mix differs by market.

The 51 state and D.C. rows sum to exactly 53.38 GW, matching the U.S. control row (NextMSC check).

Why pumped hydro is excluded

Large pumped-hydro figures can be mistaken for battery fleets. EIA's July 2026 table shows Virginia with 230.6 MW of Other Energy Storage and 3,169.0 MW of hydroelectric pumped storage, and South Carolina with 88.0 MW and 2,996.0 MW respectively. The state values used here (0.231 GW and 0.088 GW) are the Other Energy Storage figures, which is why they differ so sharply from the pumped-hydro totals.

Growth needs context

Year-on-year growth is (July 2026 capacity / July 2025 capacity) − 1. When the prior-year base is close to zero, a small addition produces a very large percentage. Utah's change is about 92,000% (roughly 921 times the prior-year level), from 1.0 MW to 921.0 MW in EIA's July 2025 and July 2026 tables, and Iowa's is 961%. Both are arithmetically consistent but say little about market scale. The chart below keeps only states with at least 0.4 GW installed and a prior-year base of at least 0.1 GW.

The Interconnection Queue: What It Is and What It Is Not

The queue is a record of projects seeking interconnection, not a construction schedule. Berkeley Lab's Queued Up 2026 edition covers requests through the end of 2025. It aggregates data from 7 ISOs/RTOs and 50 non-ISO balancing areas that together represent about 98% of installed U.S. generating capacity, and it includes only generation and storage requests seeking transmission interconnection. Load requests, distribution-connected resources and behind-the-meter projects are excluded.

At the end of 2025 about 8,200 projects were actively seeking interconnection, representing 1,312 GW of generation and approximately 749 GW of storage. Storage queue volume fell 16% from the prior year, and total active volume fell 10%. A further 549 GW already had a draft or executed interconnection agreement but had not reached commercial operation, including 161 GW of storage.

Duration and attrition

For regions with available data, the median time from interconnection request to commercial operation exceeded five years for projects built in 2025. Only 13% of capacity that submitted requests from 2000 to 2020 had reached commercial operation by the end of 2025, while 75% had been withdrawn and 10% was still active. This is a historical all-resource statistic, not a storage-specific success rate. FERC Order 2023 and related reforms are being implemented, but Berkeley Lab says it is too early to measure their full impact.

Two national queue numbers

Sources: Lawrence Berkeley National Laboratory, Queued Up: 2026 Edition; GridAlmanac state presentations; NextMSC analysis. The combined figure is not equivalent to the LBNL headline.

A solar-plus-battery request can contain a large amount of solar generation alongside its battery component, so treating the full project capacity as battery capacity overstates the battery pipeline. NextMSC therefore reports the two components separately and uses 749 GW only when citing Berkeley Lab's national storage queue.

The state rows with queue data sum to 360.20 GW of standalone battery (92.0% of the national 391.50 GW) and 388.63 GW of solar-plus-battery (90.1% of 431.55 GW). The gap reflects the 9 jurisdictions without a usable state breakout: Alaska, D.C., Hawaii, North Carolina, Oregon, South Carolina, South Dakota, Wisconsin and Wyoming.

Readiness inside the queue

Berkeley Lab's stage data give a rough readiness indicator. Of the approximately 749 GW of storage in active queues, 161 GW already has a draft or executed interconnection agreement, which is about 21.5% (NextMSC calculation).

Battery Interconnection Queue by State

Texas has the largest battery-related queue at 257.29 GW, followed by California at 103.06 GW and Arizona at 59.33 GW. Nevada reaches 57.44 GW, and Washington, Utah, Colorado, Illinois, Michigan and New Mexico also show substantial values. These are state-level presentation values and do not replace Berkeley Lab's national project-level accounting.

As a check, Texas's 257.29 GW (145.85 GW standalone plus 111.44 GW solar-plus-battery) sits below the 503.59 GW all-type queue shown for the state on GridAlmanac, and California's 103.06 GW sits below its 122.29 GW all-type total.

Standalone versus hybrid

Mix differs sharply by state. Hybrid requests make up 80% of Arizona's queue (47.53 of 59.33 GW), 75% of Nevada's, 80% of Idaho's and 75% of Utah's, which reflects solar-led development. Standalone requests make up 88% of Michigan's queue, 88% of New York's, 66% of Illinois's and 74% of Oklahoma's. Georgia is near even at 7.29 GW standalone and 7.62 GW hybrid. These shares are NextMSC calculations on the state dataset.

Queue relative to the installed fleet

The queue-to-installed ratio is battery-related queue divided by installed storage. It is a diagnostic and not a forecast. Texas is about 14.3x, California 6.4x and Arizona 8.9x, which does not mean those states will build that multiple of their fleets.

Small bases produce extreme ratios. Missouri's 7,260x reflects 7.26 GW of queue against about 1 MW installed, Washington's 6,515x reflects 26.06 GW against 4 MW, and Alabama's 4,780x reflects 4.78 GW against 1 MW. A ratio of several hundred times can come from a queue of a few GW against a tiny fleet, so it should always be read beside absolute GW.

Installed storage and queue side by side

Figure 6 compares installed storage with the battery-related queue for the eight states with the most installed capacity. A logarithmic axis is used because the two measures differ by an order of magnitude or more. Among these eight states the queue ranges from 1.85 times installed capacity in Florida to 27.5 times in Nevada, and Texas, California and Arizona sit in between at 14.3x, 6.4x and 8.9x.

The log scale preserves the order of magnitude but visually compresses the gap, so the bars should be read with the labelled values. A queue several times larger than the operating fleet is a statement about requests, not about expected additions.

How concentrated the state queue is

The nine jurisdictions without a usable breakout are excluded, so the state rows with data sum to 746.19 GW of battery-related queue. Texas alone is 34.5% of that, the top five states (Texas, California, Arizona, Nevada and Washington) are 67.4%, and the top ten are 79.7%. Nationally, solar-plus-battery requests are 52.4% of the combined 823.05 GW and standalone batteries 47.6%. Among the 40 states with both components, hybrid requests exceed standalone requests in 17 and standalone exceed hybrid in 23 (NextMSC calculations).

Modeled Storage Cost by State

There is no single observed statewide storage CAPEX figure comparable across all states. Actual project prices depend on chemistry, duration, inverter configuration, site work, interconnection scope, EPC terms, procurement timing, financing, taxes, labor and logistics. A low equipment price can still produce a high installed cost if civil or transmission work is expensive.

NREL's 2025 update is a forward-looking projection and not a database of signed contracts. It gives 4-hour lithium-ion overnight capital cost of $334/kWh in 2024 and a mid-case of $308/kWh in 2026. Its 2035 projections are $147/kWh (low), $243/kWh (mid) and $339/kWh (high) in 2024 dollars. The 2035 figures are long-term context and are not mixed with the 2026 state layer.

The state cost model

Modeled state cost = $308/kWh × Macro Cost Pressure Index / 100. The index is a NextMSC analytical screening metric normalized to 100 for the United States. It combines three state-level indicators and shows income for context only: the price-level factor is the state regional price parity divided by 100, the electricity-price factor is the state average electricity price divided by the U.S. average, and the labor-tightness factor is the U.S. unemployment rate divided by the state rate, capped at 0.75 to 1.50. The component weights are not published, because the available source inputs do not reproduce every published index value consistently, so the published index values are used as given. The cost step itself is exact on all 51 state rows. California, for example, has an index of 112.5 and a proxy of $346.50/kWh. That does not mean California projects transact at $346.50/kWh; it is a reproducible comparison tool.

The proxy ranges from $266.73/kWh in Arkansas to $347.42/kWh in D.C., with California ($346.50), Hawaii ($341.26), Washington ($338.18) and Massachusetts ($336.95) at the high end. Texas is $300.30/kWh, below the U.S. base because its index is 97.5. That is a model output and not an observed price discount.

What drives realized cost

Battery pack cost is only one part of a utility-scale system. NREL's bottom-up framework includes the battery pack, inverter and balance-of-system components and installation costs, and duration changes the balance between energy and power costs. Land, grading, local labor, permitting, taxes, interconnection scope and delivery logistics add project-level differences that a national battery price cannot capture. Conversely, higher local cost pressure does not guarantee higher realized CAPEX if a developer has a favorable equipment contract or a standardized design.

Illustrative cost arithmetic

To show what the proxy implies at project scale, the table applies each modeled $/kWh figure to a hypothetical 100 MW, 4-hour (400 MWh) system.

Source: NextMSC arithmetic on the modeled proxies (400,000 kWh x $/kWh). Illustrative only.

The spread between the lowest and highest proxy is about $31.9 million on a system of that size, or about 30%, which shows why a single national figure can mislead. NREL's own path adds time context: its mid-case falls from $334/kWh in 2024 to $308/kWh in 2026, a decline of about 7.8%, and to $243/kWh in 2035, about 21% below the 2026 level (NextMSC calculations). Those are national projections in 2024 dollars and are separate from the state layer.

Regional Profiles: California, Texas, the West, Midwest and Northeast

California

California combines a large installed base with one of the largest queues: 16.18 GW installed, 45.61 GW of standalone battery requests and 57.45 GW of solar-plus-battery requests, for 103.06 GW, about 6.4 times the fleet. Installed growth was 21.1%, the slowest of the large markets, consistent with an established fleet. The modeled cost proxy is $346.50/kWh.

Texas

Texas is the largest installed market at 18.02 GW (+72.5% year on year) and has the largest queue at 257.29 GW, split 145.85 GW standalone and 111.44 GW hybrid, a ratio of about 14.3x. A large queue can sit alongside a more modest build rate when transmission, project economics, equipment availability and interconnection studies constrain conversion. The modeled cost proxy is $300.30/kWh.

Arizona, Nevada and the West

Arizona has 6.65 GW installed (+60.7%) and 59.33 GW of queue, 80% of it hybrid. Nevada has 2.09 GW installed and 57.44 GW of queue, New Mexico 1.36 GW and 15.76 GW, Utah 0.92 GW and 23.66 GW, and Idaho 0.66 GW and 15.19 GW. The Western pattern is storage developed alongside solar. The data do not support attributing it to one policy or market feature.

Midwest and Northeast

Michigan has 0.12 GW installed and 16.10 GW of queue, and Illinois 0.10 GW and 17.67 GW. New York has 0.28 GW and 12.43 GW, and Massachusetts 0.74 GW and 5.17 GW. These states have small operating fleets but meaningful queues, which may mean an early deployment stage or projects at very different levels of readiness. Queue size alone cannot distinguish them.

Regional Divisions and Where the Additions Came From

EIA's Table 6.2.A also groups states into Census divisions, which shows where the 17.9 GW added between July 2025 and July 2026 landed. The U.S. Other Energy Storage total rose from 35,513.0 MW to 53,380.5 MW, an addition of 17,867.5 MW. The division figures below are EIA's; the additions, growth rates and shares are NextMSC calculations.

Source: U.S. EIA, Electric Power Monthly Table 6.2.A (July 2026 and 2025). Added, growth and share columns are NextMSC calculations. Values are preliminary.

West South Central, which includes Texas, held 34.6% of the fleet and took 43.8% of the additions. Pacific Contiguous held 31.3% of the fleet but took only 15.8% of the additions, because its base was already large. The Mountain division grew the fastest of the large regions, by 68.4%, adding 5,012.2 MW. South Atlantic grew 97.5% from a much smaller base.

Where the additions came from

Additions were more concentrated than the fleet itself. Texas alone added 7,574.8 MW, 42.4% of the national increase. Texas, California and Arizona together added 12,903.5 MW, or 72.2% of the total, and the ten largest additions account for 91.2%.

The national 50.3% growth rate is therefore not broad-based: most of the increase came from a handful of states, the same ones that dominate the installed fleet and the queue.

Storage as a Share of State Capacity

Another way to read the fleet is relative to the size of each state's power system. EIA's Table 6.2.A reports All Sources net summer capacity alongside Other Energy Storage, so dividing one by the other gives storage as a share of reported capacity. Nationally the figure is 4.06% (53,380.5 MW of 1,315,876.2 MW).

Source: U.S. EIA, Electric Power Monthly Table 6.2.A, July 2026. Share is a NextMSC calculation.

Arizona (16.0%), California (15.6%) and Hawaii (14.8%) have the largest storage shares, ahead of Nevada (11.7%), Idaho (10.0%) and Texas (9.3%). Texas has the largest fleet in absolute terms but a smaller share than California or Arizona because its total capacity base is almost twice as large as California's. Florida's 1,199.6 MW is only 1.6% of its 75,067.4 MW, and Colorado's 666.3 MW is 3.1% of its capacity.

The ranking differs from the ranking by absolute GW: Hawaii and Idaho appear because they have modest fleets on small systems. The share view shows where storage is large relative to the grid it serves, which is a different question from where the largest fleets are.

Why State Storage Pipelines Differ

State pipelines differ because several factors interact:

• Renewable penetration and curtailment exposure, which raise the value of shifting energy.
• Capacity and resource-adequacy requirements.
• Wholesale energy and ancillary-service market design.
• Utility procurement and state policy mechanisms.
• Transmission availability and interconnection upgrade requirements.
• Land, labor, permitting, taxes and construction conditions.
• Battery equipment pricing, duration and project configuration.

The dataset does not test which of these matters most in any state. It supplies the quantities against which those explanations can be checked.

State Clusters: How to Read the Dataset

The groups below are descriptive and are not formal market tiers or rankings.

Source: NextMSC state dataset (EIA installed capacity; GridAlmanac/LBNL-derived queue).

State Data Quality and Edge Cases

Source: NextMSC state dataset.

A zero in the installed series reflects the category and coverage used here, not necessarily the absence of storage in every technology or behind-the-meter segment. The article concerns the utility-scale source series. No missing value is filled from a neighboring technology, and no modeled value is described as observed.

Dataset Explorer

Filter the state-level installed capacity, queue and modeled cost data compiled for this analysis. N/A means no defensible state-level figure in the cited coverage. The U.S. control row is pinned at the top.

Sources: U.S. EIA (Table 6.2.A, July 2026), Lawrence Berkeley National Laboratory (Queued Up 2026), GridAlmanac state queue presentations derived from LBNL, NREL (2025 cost update) and NextMSC analysis. Queue values are requested interconnection capacity, not expected buildout. Ratio, index and modeled cost columns are NextMSC calculations or modeled values.

Methodology, Definitions and Limitations

Data cut-off: September 30, 2026. Installed capacity is the July 2026 snapshot, queue data run through December 31, 2025, and the cost layer is a modeled 4-hour proxy. Because the series have different dates, observation dates are shown beside the metrics. Planned refresh: quarterly, preserving field definitions.

Source hierarchy

Installed capacity uses EIA Electric Power Monthly Table 6.2.A for July 2026 and July 2025. NextMSC checked all 51 state values and the U.S. total (53,380.5 MW against 35,513.0 MW) against that table and found no differences. Queue data use GridAlmanac state presentations, which are derived from LBNL data, where a reproducible state breakout exists, with Berkeley Lab's Queued Up 2026 edition as the national anchor. Cost assumptions use NREL's Cost Projections for Utility-Scale Battery Storage: 2025 Update.

Data types

Reported: presented directly by the source under its own definition. Calculated: derived arithmetically from reported values, such as YoY growth and queue-to-installed ratio. Modeled: generated from the NextMSC Macro Cost Pressure Index and the NREL-based U.S. base.

Formulas

YoY growth = (July 2026 capacity / July 2025 capacity) − 1. Battery-related queue = standalone battery queue + solar-plus-battery primary-component queue, where both are available. Queue-to-installed = battery-related queue / installed storage. Modeled state cost = $308/kWh × index / 100. The index values are published as given, without component weights, and the income factor is context only.

Limitations

• The EIA category is “Other Energy Storage”, not a chemistry-specific battery census, and EIA labels the values preliminary.
• LBNL queue data are transmission-focused and exclude distribution-connected and behind-the-meter resources.
• State queue presentations differ in category visibility and may require visible project-list sums.
• Queue capacity is not a forecast of construction or commercial operation.
• Modeled costs are not observed project prices and are not procurement quotes.
• The 749 GW LBNL storage queue and the 823.05 GW combined state view are not interchangeable.

Update protocol

Future editions should refresh the EIA snapshot and keep the prior-year snapshot for YoY, refresh the queue data to the latest common cutoff, reconcile national totals to the LBNL headline before updating the state combined field, update the NREL base only when a new annual projection is released, use the published index values as the state inputs (revising index values and methodology together if the model changes), and run denominator checks on growth and ratio outliers. Articles, charts and tables should be generated from the master dataset so that a correction propagates everywhere.

Frequently Asked Questions

How much battery storage does the U.S. have?

EIA reports 53.38 GW of utility-scale storage in its Other Energy Storage category for July 2026, up 50.3% from about 35.51 GW a year earlier. Texas (18.02 GW) and California (16.18 GW) hold about 64%.

How big is the battery interconnection queue?

Berkeley Lab reports approximately 749 GW of storage in active transmission queues at the end of 2025. NextMSC's state dataset shows a different construct, 823.05 GW, that adds standalone battery requests to the primary capacity of solar-plus-battery requests.

Will the queue be built?

Not in full. Only 13% of capacity that submitted requests from 2000 to 2020 had reached commercial operation by the end of 2025, and 75% had been withdrawn. That statistic covers all resource types, not storage alone.

What does battery storage cost by state?

No observed statewide series exists in the sources used. NextMSC models a 4-hour proxy by scaling NREL's $308/kWh 2026 mid-case by a state index, giving $266.73/kWh (Arkansas) to $347.42/kWh (D.C.). These are comparison indicators, not project prices.

Why are some state ratios in the thousands?

Because the installed base is close to zero. Missouri's 7,260x reflects about 1 MW installed against 7.26 GW of requests.

Which states added the most storage in the past year?

EIA's July 2025 and July 2026 tables show Texas adding 7,574.8 MW, California 2,817.7 MW and Arizona 2,511.0 MW, followed by Utah (920.0 MW) and New Mexico (671.2 MW). Together the top three account for about 72% of the national increase.

How large is storage compared with the rest of a state's capacity?

Nationally, Other Energy Storage is about 4.1% of EIA's reported capacity. Arizona (16.0%), California (15.6%) and Hawaii (14.8%) are the highest, and Texas is 9.3%.

What the Data Shows and the 2026–2030 Outlook

The data support five durable observations. First, utility-scale storage is concentrated in a small number of states. Second, the pipeline is far larger than the operating fleet. Third, hybrid solar-plus-battery requests are a substantial part of that pipeline and must be separated from standalone batteries. Fourth, queue duration and attrition mean pipeline capacity is a development signal and not expected supply. Fifth, state cost comparisons require a model when no observed project-price dataset exists.

For market researchers, “capacity” should always be qualified as operating MW, energy capacity in MWh, requested interconnection MW or modeled cost capacity. For investors and developers, the dataset is a screening framework showing where installed assets, development interest and cost pressure intersect. Project underwriting still requires site-level interconnection, procurement and revenue analysis.

The 2026 to 2030 outlook depends on the interaction of demand growth, renewable deployment, reliability requirements, market design and interconnection reform, and not on one national growth rate. The large queue is evidence of interest, and Berkeley Lab's attrition statistics are a reminder that it will shrink substantially before reaching operation. NREL's projections point to continued long-term cost declines, but realized state costs can move differently because site and construction conditions are not uniform. A scenario-based approach is more appropriate than a single cost curve.

What to monitor