Why Batteries Change Fast Beyond Economy

For most of the last decade, the story of grid-scale battery storage was a story of one direction: costs down, capacity up, chemistry settling. Lithium iron phosphate won the stationary market, container sizes standardised, and a handful of Chinese manufacturers took commanding share. It was reasonable to assume the hardware was maturing toward a stable plateau, and to design software on that assumption.

That assumption is breaking. Battery storage costs have fallen by roughly two-thirds since 2022, and analysts now openly debate whether the floor is close or whether a new chemistry is about to push it lower again. The dominant supplier's market share has slipped materially as the field fragments. A chemistry that was a laboratory curiosity five years ago is entering commercial grid deployment. The physical container is being redesigned around bigger cells and taller stacks. And a duration limit everyone treated as fixed has already been broken. For anyone building or buying the control systems that run these assets, the lesson is uncomfortable: the hardware underneath the software is changing faster than the software industry planned for, and the pace is accelerating.

This piece works through what is actually changing — in cost, in chemistry, in suppliers, in container design, in duration — and then asks the question the industry has largely avoided: if the battery beneath the control layer is going to keep changing every couple of years, what does that mean for the software layer that has to run it?

The Cost Collapse, and Why It Might Reverse

The headline number is stark. Battery storage costs have fallen by about two-thirds since 2022, driven by falling raw-material prices, manufacturing scale, and fierce competition among cell makers. Chinese cell prices have pushed toward a widely-discussed target of$50 per kilowatt-hour, a level that would have seemed implausible only a few years ago. Cheaper storage has reshaped project economics across every application, from energy arbitrage to renewable firming.

What makes the current moment interesting is that the fall may not continue smoothly, and could even reverse. Much of the recent decline was driven by unusually low lithium carbonate prices, which cannot be relied on to stay depressed indefinitely; a rebound in raw-material costs would push cell prices back up. At the same time, the industry is approaching the point where the easy gains from squeezing existing chemistry are exhausted. The next leg down, if it comes, will not come from optimising lithium iron phosphate further. It will come from replacing it.

Sodium-Ion: The Chemistry That Could Repeat LFP's Takeover

Lithium iron phosphate's rise offers the template for what may be about to happen again. LFP went from a minor chemistry to the dominant grid-storage technology in only a few years, displacing the nickel-based chemistries that preceded it, because it was cheaper, safer, and longer-lived even though it gave up some energy density. The market reorganised around it faster than most forecasts predicted.

Sodium-ion is positioned to follow the same path in cost-sensitive stationary storage. It uses sodium, which is roughly a thousand times more abundant in the earth's crust than lithium, and it can be built without cobalt, nickel, or lithium at all. Early commercial cells are reaching cost levels meaningfully below lithium iron phosphate, and they bring genuine advantages for grid use: strong cold-weather performance, high cycle life, and improved thermal stability that eases the fire-safety burden. Recent commercial grid-storage cells have been announced with cycle lives well beyond what LFP delivers and capacity retention at temperatures where lithium systems need active heating.

Sodium-ion will not replace lithium everywhere; its lower energy density keeps it out of applications where space and weight are tight. But grid storage is precisely the application where its weaknesses matter least and its cost and abundance advantages matter most. The likely outcome is not a single winner but a bifurcated market where sodium-ion takes a large share of cost-sensitive stationary storage while lithium retains the premium and space-constrained roles. For the control layer, a bifurcated market is arguably harder than a clean handover, because it means runningboth chemistries, often across one portfolio, at the same time.

Why Chemistry Change Is a Software Problem, Not Just a Hardware One

Here is the part the cost-and-chemistry coverage consistently omits. Different chemistries do not behave the same way electrically, and the software that estimates their state cannot simply be copied across. Sodium-ion and lithium iron phosphate have different open-circuit-voltage curves, different hysteresis, and different internal resistance behaviour across temperature and state of charge. Research on migrating battery-management logic from lithium to sodium has found that applying lithium estimation algorithms directly to sodium cellscompromises state-estimation accuracy, with one study measuring peak voltage errors above 25% before the algorithms were re-tuned for the new chemistry. Estimating state of charge and cycle life accurately is chemistry-specific work, not a universal formula.

This is why chemistry churn lands on the software layer with real force. Every time the cell underneath changes, the assumptions baked into state-of-charge estimation, degradation modelling, thermal limits, and dispatch optimisation have to be revisited. A control system built rigidly around the electrical behaviour of one chemistry does not gracefully accept another. The systems that will cope are the ones designed from the start to be chemistry-aware and reconfigurable rather than hard-wired to lithium iron phosphate's particular quirks.

The Supplier Market Is Fragmenting, Not Consolidating

Conventional wisdom held that battery manufacturing would consolidate around a few dominant giants as scale advantages compounded. The opposite is happening. The largest manufacturer's share of the global energy-storage battery market has fallen substantially from its peak as a widening field of capable competitors takes ground, and market-share tables now shuffle from one reporting period to the next. The names that dominate today are not guaranteed to dominate in five years.

For asset owners and developers, fragmentation is mostly good news: more suppliers means more competition, more price pressure, and more negotiating leverage. But it carries an integration cost that is easy to underestimate. A fragmenting supplier market means the batteries, power conversion systems, and components a developer procures are increasingly likely to come from different manufacturers, chosen on price and availability at the moment of purchase rather than dictated by a single-vendor relationship. That only works if the control layer can integrate whatever combination the procurement process produces.

This is where a hardware-agnostic, vendor-neutral control platform stops being a marketing phrase and becomes a procurement precondition. A system locked to one manufacturer's batteries forces every future purchase back to that manufacturer, forfeiting exactly the competition and flexibility that a fragmenting market offers. A vendor-neutral platform lets a developer buy the best available cell each time, from whichever supplier wins on merit, and integrate it without re-platforming. The distinction between the control layers that manage this and those that do not is explored in the guide to EMS, BMS, SCADA, and PPC.

Container Design Is Changing Again: Bigger Cells, Taller Stacks

The physical form of grid storage is not standing still either. For years the trend was toward larger container capacities packed with ever-bigger cells, cramming more energy into the same footprint. That direction is now shifting: rather than only enlarging cells, manufacturers are redesigning the container itself, moving toward taller stacks and reconfigured internal architectures to raise energy density per unit of land.

This matters for the control layer because container architecture affects how cells are grouped, monitored, and thermally managed. A taller-stacked, higher-density container has different thermal gradients and different cell-balancing demands than the wide, low containers that preceded it. Managing heat well becomes more important as density rises, since thermal stress is a primary driver of degradation, a relationship covered in the guide to how operation affects battery health and lifetime. Each redesign of the physical container quietly rewrites some of the assumptions the control system depends on.

Density is not a free win, either. Packing more energy into a smaller footprint raises the fire-safety stakes, making detection, isolation, and prevention more critical rather than less. The evolving safety demands of denser systems are covered in the guide to BESS safety and thermal runaway.

The Duration Limit That Everyone Assumed Was Fixed

Perhaps the clearest sign that the ground is moving is duration. For years, grid storage was effectively synonymous with short-duration systems, and four hours became the de facto ceiling for lithium projects on cost grounds. Longer durations were assumed to belong to other technologies entirely. That assumption has broken: lithium systems of sixteen to eighteen hours have been deployed, shattering the duration limit the industry treated as fixed and blurring the line between conventional storage and long-duration storage.

This shift is being reinforced by policy. Long-duration energy storage support schemes, such as the mechanism developed by the British regulator Ofgem, are creating deliberate market pull for systems that can discharge over far longer windows, alongside a wave of alternative long-duration technologies — flow batteries, iron-air, and even compressed-CO2 storage — competing for the same role. Whichever technologies win, the duration envelope the control layer must manage is widening dramatically, from sub-hour frequency response to multi-day discharge. The way a system is sized around power, energy, and duration is set out in the guide to how to size a BESS, and the calculus is changing as the duration ceiling lifts.

What This Means for the Control Layer

Step back from the individual shifts and a single pattern emerges. Cost, chemistry, suppliers, container design, and duration are all moving at once, and they are moving faster than the roughly decade-long horizon on which grid infrastructure is usually planned. A battery bought today may be a different chemistry, from a different supplier, in a different container, than the one bought for the same site three years from now.

This reverses the intuition that the battery is the asset and the software is an accessory. In a fast-churning hardware market, the control and optimisation layer is the element that persists across successive generations of cells, containers, and suppliers, and it is the element that has to absorb every change beneath it. That places a premium on three properties that were once considered nice-to-have and are becoming essential.

Chemistry-agnostic by design.A control layer that treats lithium iron phosphate's electrical behaviour as universal will misestimate the state of a sodium-ion pack. One built to be reconfigured for the chemistry it is actually running — its voltage curve, its hysteresis, its thermal limits, its degradation profile — can adopt a new chemistry without being rebuilt. As chemistries multiply, this becomes the difference between a platform that lasts and one that is obsoleted by the next cost breakthrough.

Vendor-neutral by architecture.A fragmenting supplier market only benefits the buyer if the control layer can integrate any supplier's hardware. Genuine vendor neutrality — integrating batteries, power conversion systems, and components from any manufacturer through open protocols — is what lets a developer keep buying the best available cell rather than being funnelled back to a single vendor by their software. This is the same principle that makes mixed-vendor and mixed-vintage fleets manageable, as in second-life and repurposed-battery deployments.

Forward-compatible in duration and application.A platform that assumed four-hour lithium duration is poorly placed to run a sixteen-hour system or coordinate a long-duration technology alongside conventional storage. One built to manage the full envelope — from sub-second frequency response to multi-day discharge, across microgrids, data-centre power, and grid-scale portfolios — does not have to be replaced each time the application envelope expands.

There is a service and warranty dimension to this as well. As suppliers proliferate and chemistries turn over, the spare-parts and warranty landscape grows more complex, and the operating data that a control layer captures becomes the record that protects warranty claims and residual value across a hardware base that is no longer uniform. The software is not only running the asset; it is the continuity layer that holds a heterogeneous, fast-changing fleet together.

The Takeaway: Standardise on the Layer That Lasts

The batteries are going to keep changing. Sodium-ion may do to lithium iron phosphate what lithium iron phosphate did to nickel chemistries. Costs may fall further or rebound on raw-material prices. Suppliers will keep trading places. Containers will keep being redesigned. Durations will keep stretching. None of this is a reason to wait, because there is no stable end-state to wait for — the churn is the steady state now.

The strategic response is not to bet on a single chemistry, supplier, or format, but to standardise on the one layer that can absorb all of them: a control and optimisation platform that is chemistry-agnostic, vendor-neutral, and forward-compatible by design. In a market where the hardware is changing faster than anyone expected, the software that does not have to change with it is the closest thing to a durable asset. The developers who internalise this will keep their freedom to buy the best hardware available at every turn. The ones who lock their software to today's winning battery will find themselves re-platforming the moment the next one arrives — which, on current evidence, will be sooner than they think.