Grid Battery Storage Sets US Records as Carbon Capture Slips Behind

Batteries are scaling while carbon removal waits

Mid-October 2026 coverage of climate technology points one way. Technologies that store electricity and move it around the grid are growing quickly. Technologies that pull carbon out of the air are still trying to show they can run at the scale they were built for. Across storage installations, battery chemistry, startup funding and carbon capture, the deployable, revenue-earning parts of the sector are getting the money. Carbon removal is getting a smaller share and more modest goals.

A record storage quarter, measured two ways

The clearest number is US battery deployment. The American Clean Power Association (ACP) and Wood Mackenzie report that the country installed a record 18.9 GWh of battery storage in the second quarter of 2026, up 17% from a year earlier.37 The Solar Energy Industries Association (SEIA) and Benchmark Mineral Intelligence put the same quarter at 20.2 GWh. They call it the largest storage quarter in US history, and say Arizona alone added 6.2 GWh, a record for any single state.31

The two figures differ because they come from different data series. Both still describe the strongest quarter the US has ever had. The ACP report adds an important detail: installed power capacity actually fell 7% to 5.4 GW. Average system duration rose from 2.8 hours to 3.5 hours over the year.36 One analysis argues the record is a record for stored energy, not for power. It links the change to resource adequacy rules that reward longer discharge, which reflects how utilities buy storage more than any change in cell costs.34

That matters. Grid operators are no longer paying only for short bursts of power at the evening peak. They want batteries that can cover the whole peak. Utility-scale capacity fell 8% in megawatt terms, which the report blames on saturation in leading markets. Wood Mackenzie expects full-year 2026 installations to roughly match 2025.40 Average utility-scale system prices fell 2% to $916 per kW.37

The bigger picture is still strong growth. Developers plan to add 24 GW of utility-scale batteries in 2026, up from a record 15 GW in 2025. Texas accounts for 53% of the planned additions.33 Batteries make up 28% of planned US capacity additions this year, behind only solar.39 SEIA-linked reporting says US utility-scale storage nearly doubled from 88 GWh to 165 GWh over 18 months. It also says batteries supplied more electricity in the first eight months of 2026 than in all of 2025.31 California's fleet has grown from under 700 MW in early 2019 to more than 23,000 MW by mid-2026, according to state energy officials.32

Home batteries act like a power plant

Some storage is not built at grid scale at all. Sunrun and Tesla say their combined home batteries delivered 745 MW of peak power to California's grid on October 6. The day before, they delivered 669 MW. The previous record was 580 MW, set on September 9.35 The dispatches were triggered when day-ahead wholesale prices passed $200 per megawatt-hour. Sunrun says it has more than 120,000 home batteries enrolled in California.35

The company's own framing is that home batteries can compete with large power plants. Even setting the marketing aside, four records in about 15 months show networks of home batteries becoming a real tool for managing peaks. A Google-funded project with PG&E in the San Jose area is due to start dispatching this month. It will test whether these devices can also relieve strain on local parts of the grid.35

Sodium-ion moves from labs to factories

The main battery-chemistry story of 2026 is sodium-ion. For grid storage, lithium iron phosphate (LFP) still dominates, at about 90% of deployments, up from well under half five years ago.33 The pressure on LFP is coming from sodium.

Reports say CATL and HyperStrong signed a 60 GWh sodium-ion supply deal in April, the largest such agreement announced. BYD is reported to be ramping up a 30 GWh sodium-ion plant in Xuzhou.29 CATL says its Naxtra sodium-ion line has reached production at the scale of gigawatt-hours, with full mass production due by the end of 2026. On September 9, the engineering standards body SAE International issued J3335, a set of test procedures for sodium-ion cells that power electric vehicles.30 Standards bodies usually write rules like this only once they expect a technology to reach the market.

Independent testing supports the trend. A teardown of 120 sodium-ion cells from Chinese maker Hina found them unusually uniform. Their tabless design, which reduces electrical resistance, resembles Tesla's architecture.24 The same researchers found clear weaknesses, though. Charging in cold weather is poor, energy density still trails leading lithium-ion cells, and copper was unevenly spread through parts of the cathode.24

In the US, Peak Energy picked Sacramento for what is described as the country's first factory for grid-scale sodium-ion systems. The $71 million plant is meant to produce 4 GWh a year from early 2027.29 Coverage also notes that Western companies trail China in sodium-ion manufacturing, and that Chinese capacity already far exceeds demand.26

Lab results keep arriving. University of Surrey researchers report that leaving water inside a sodium vanadate cathode material, instead of removing it during manufacturing, nearly doubled the charge it could store over more than 400 cycles.27 The researchers themselves say the work is at an early stage. That fits a pattern across battery science: an improvement in one component does not finish the job, because the whole cell has to improve together.28

Carbon capture improves its materials as policy support shrinks

Carbon removal tells a harder story. Georgia Tech and CarbonCapture Inc. report a one-step treatment for a standard commercial resin used to capture CO₂. It boosts capture by about 75% in dry, open-air conditions and roughly halves breakdown from oxidation.11 The team points to the US Department of Energy's target of capturing and storing CO₂ for under $100 per net tonne.11 Climeworks also reported upgrades at its Mammoth plant in Iceland. It says capture performance more than doubled in the upgraded units while operating costs fell by more than half.15

Actual output is far more modest. Mammoth is built for 36,000 tonnes a year but removed 675 net tonnes in the first half of 2026, about 4% of capacity.12 Occidental's 500,000-tonne Stratos plant in Texas has slipped to a target start-up at the end of 2026. Microsoft, the dominant buyer of carbon removal, reportedly paused new purchases in April.12 Funding data shows how concentrated demand was. Microsoft accounted for more than 80% of named-buyer forward contracts for long-lasting carbon removal.19

Policy coverage disagrees on the numbers. One market report still describes more than $3.5 billion in DOE commitments to air-capture hubs.15 Other reporting says the hub program was cut to about $1.2 billion between October 2025 and April 2026. It notes that the 45Q tax credit, worth $180 per tonne for direct air capture with dedicated storage, survived the 2025 budget law.19 Our reading is that the higher figure is out of date. The industrial version of carbon capture, which captures CO₂ from factory exhaust, kept its subsidy. Removal directly from the air lost much of its federal support.12

What remains is smaller and more cautious. Dutch firm Skytree got a €1 million loan to prepare a geothermal-powered site in Iceland, with machines due in 2028.18 Canada's Deep Sky is counting on rules for trading credits internationally under Article 6 of the Paris Agreement, and on CORSIA, the international aviation offset scheme. CORSIA's mandatory phase starts in January 2027.16

Where startup money is going

Venture funding reflects the same priorities. In 12 recent climate deals worth about $553.9 million, three companies took roughly 65% of the money. Reverion's $175 million Series B was the largest. The smallest was a $700,000 raise for a carbon-capture startup.1 Demand-response company Voltus raised a $225 million Series B+.8 France's Epyr raised €15 million for industrial heat storage after its first commercial system began supplying up to 25% of a paper mill's steam.6 One recent summary put it bluntly: climate money has moved toward whatever powers AI data centers.3

Our reading

The dividing line is between assets that earn revenue today and assets that depend on someone choosing to buy them. Batteries, home-battery networks and thermal storage are paid by electricity markets that need flexibility as demand from AI rises. Sodium-ion has entered that market through factories and standards, not only lab papers. Direct air capture, by contrast, depends on a few voluntary buyers and on uncertain federal support. In that position, real improvements in materials do not yet turn into tonnes removed. If sodium-ion meets its 2027 production targets, the gap between storage and air capture is likely to grow. Air capture's next test is simpler: whether Stratos runs on schedule.