Explained, Technology
Battery Storage
Grid-scale battery storage is the youngest technology on this platform and the fastest-growing power technology in history: an installed base eleven times larger than it was in 2021, and past pumped storage in power terms as of 2025. It is also the most misunderstood, because a battery is not a power station. It generates nothing. Its entire value is temporal, buying electricity when the system has too much and selling it back when the system has too little, and its entire character is set by software deciding, second by second, which of a dozen jobs is worth doing right now.
The system
A grid battery is cells in modules, modules in racks, racks in containers, containers behind inverters, and the whole thing under three nested layers of software. The hardware is increasingly a commodity. The stack is the product.
Lithium iron phosphate, LFP, has won the grid: cheaper than the nickel chemistries in vehicles, longer cycle life, and markedly more tolerant of abuse. Modern grid cells are large prismatic units, and a standard shipping-container-sized enclosure now packs around five megawatt-hours, a figure that has roughly doubled in a few years without the container getting any bigger.
The lowest software layer, watching every cell's voltage and temperature, balancing charge across thousands of cells, and enforcing the limits that keep chemistry as chemistry rather than as an exotherm. The BMS is simultaneously a performance system, a warranty instrument and a safety system, and it never sleeps.
Inverters converting the DC battery to grid AC in both directions, with the same grid-code duties as a solar inverter plus one more: increasingly, grid-forming operation, holding voltage and frequency the way heavy spinning machines once did. A battery that can do this is no longer just storage, it is grid stability as a service.
The top layer, deciding what the plant does: charge on this price, hold reserve for that contract, respond to frequency within milliseconds. The EMS is where revenue is made, usually run or co-run by an optimiser trading the asset across several markets at once. Two identical batteries with different software earn materially different money.
What it earns, and the duration question
A battery stacks revenues: energy arbitrage across the daily price spread, frequency response, capacity payments, constraint management, and, co-located with solar or wind, absorbing the output that would otherwise be curtailed. The mix shifts as fleets grow, the frequency markets that made early projects rich are shallow and saturate quickly, leaving arbitrage, which solar's midday glut keeps widening, as the durable core.
Power and energy are different quantities, and the distinction runs the whole sector. A plant rated 100 megawatts with two hours of duration holds 200 megawatt-hours; it can shift lunch to dinner but not summer to winter. Durations are lengthening, two hours towards four and beyond, as the daily spread deepens, pushing batteries into territory once reserved for pumped hydro. The two are complements more than rivals: batteries shift hours with unmatched speed and siting freedom, pumped storage shifts days at a scale no battery fleet approaches.
What breaks, and what burns
The honest section. Battery failures are mostly boring, and the exceptions are the industry's defining risk.
- Degradation. Not a failure but the design constraint: every cycle costs capacity, so projects are financed around a warranted degradation curve and routinely "augmented", new racks added mid-life, to hold the contracted energy. Cycling strategy is a commercial decision with a chemical price.
- Power conversion and cooling. The workaday fault categories: inverter trips, HVAC failures in a machine whose chemistry cares about temperature, and the connector and contactor faults of any large DC plant.
- Thermal runaway. The one that matters. A cell driven into failure, by defect, damage or abuse, heats its neighbours, and an unchecked cascade is a fire that water does not easily stop and that can reignite days later. It is rare, well understood, and unforgiving of design shortcuts. The January 2025 Moss Landing fire in California, which destroyed much of one of the world's largest installations, an older, building-housed, nickel-chemistry design, reset public and regulatory scrutiny overnight.
- The mitigations that work. Modern practice is containerised units with distance between them, LFP's higher abuse tolerance, cell-level monitoring, off-gas detection, deflagration venting, and letting a burning container burn itself out while protecting its neighbours. Standards, NFPA 855 and its international counterparts, now encode most of this. The industry's fire record per gigawatt has improved markedly as the fleet has grown; it will never be zero, and honest operators say so.
Control systems, and keeping them safe
A battery plant is the most software-defined asset on the grid: unmanned, remotely operated, remotely traded, remotely updated, with vendor connections into the BMS, an optimiser connected to the EMS, and market interfaces on top. It is also the only asset on this platform where the cyber question and the safety question are the same question. The layers that make money and the layers that prevent thermal runaway share networks and, sometimes, vendors; a corrupted or malicious command path is therefore not just a revenue event. Every serious standard, and every serious operator, treats the protection layer as something that must fail safe independently of everything above it, precisely so that no instruction from the clever layers can talk the plant out of its limits.
The estate-level discipline is the familiar one, segmentation, controlled remote access, monitored OT traffic, firmware provenance, applied to a fleet that is young, growing at 40 percent a year, and largely built by teams for whom speed to market was the selection pressure. The machinery of doing it properly is explained at OT Cybersecurity.
Where it fits in the system
Batteries are the transition's shock absorber: the answer to solar's midday glut, the fastest frequency response on any grid, and the first tool reached for wherever curtailment signals a grid that cannot absorb what the fleet generates. Co-location with solar is becoming the default project shape, and offshore wind operators are beginning to pair coastal batteries with their export constraints for the same reason. What batteries do not do is seasons; that remains pumped storage's territory, covered on its own page. The market picture across all six technologies is on Global Markets.