A lithium-ion battery pack in 2025 costs roughly $108/kWh at the global volume-weighted average, according to BloombergNEF's annual price survey published in December — an 8% decline from $115/kWh a year earlier and the steepest fall since 2017. That figure, however, describes a DC block leaving a factory gate. It is not a price any developer, EPC or homeowner actually pays, and it does not explain where the money goes. Strip a pack down and the cell accounts for 60-75% of its cost, with the remaining quarter distributed across a long tail of electronics, thermal hardware, structural parts, testing, freight and warranty reserve. Understanding that tail is what separates a credible $/kWh quotation from an optimistic one.
The headline number also conceals enormous regional spread. BNEF's 2025 survey recorded an average pack price of $84/kWh in China, with US and European packs 40-60% higher for the same chemistry, driven by labour, energy, tariffs, lower line utilisation and a thinner domestic supply chain. For anyone procuring storage in 2026, the useful question is not "what does a battery cost" but "which of the twelve line items in this cost stack am I actually paying for, and which have been netted out."

The cell still dominates, but its share is shifting

Cell production remains the single largest cost centre in any pack. In a Chinese-made LFP system, prismatic cells typically land at $55-65/kWh at the DC block level, representing 65-75% of a pack that sells for $85/kWh. The ratio is higher for LFP than for nickel-rich chemistries because LFP cells are cheap to make relative to the hardware that surrounds them — there is less cell cost to dilute the BMS, contactors and enclosure. As cell prices fall faster than balance-of-plant costs, the non-cell share of the pack rises mechanically.
This has a practical consequence for procurement teams. When cell prices dropped roughly 60% between 2022 and 2025, pack prices did not follow at the same rate, because a BMS master unit, a set of high-voltage contactors and a UL-listed enclosure cost broadly the same in 2025 as they did in 2022. BNEF has repeatedly flagged that the balance-of-pack cost is becoming the binding constraint on further price declines in markets outside China, where assembly labour, certification testing and freight add proportionally more.
Where the cell content is cheap enough — as in Chinese utility-scale DC blocks, which traded at times below $50/kWh in 2025 tenders — the surrounding hardware can account for nearly 40% of the delivered price. That inversion is why rack-level integration, not cell chemistry, is now the main competitive battleground among system vendors, including the rack and cabinet lines catalogued on DLXN's lithium battery pages.

Inside the cell: a materials bill of materials

At cell level, the cost stack for LFP is dominated by four inputs: cathode active material, graphite anode, electrolyte and separator. Cathode powder — lithium iron phosphate — runs roughly $8-12/kWh of cell capacity in 2025, graphite anode around $5-7/kWh, electrolyte $4-6/kWh and separator $2-3/kWh. Foils, binder, conductive carbon and tab materials add another $3-4/kWh. The remainder, roughly $18-22/kWh, is manufacturing: coating, calendering, slitting, stacking or winding, electrolyte filling, formation, ageing, drying-room energy, depreciation and yield loss.
Yield deserves separate mention because it is where cost models most often go wrong. A mature Chinese LFP line runs first-pass yields above 96-98% on prismatic formats; a new entrant ramping a 300 Ah cell may sit at 85-90% for several quarters. Each percentage point of scrap is charged twice — once for the lost materials and once for the wasted capacity — and can add $2-4/kWh to cell cost during ramp. This is a principal reason why announced Western gigafactory capacity and realised low-cost output have diverged so sharply since 2023.
Materials costs have also reset violently. Lithium carbonate, the reference commodity for the whole chain, traded above RMB 600,000 per tonne in November 2022; through 2025 it has hovered in a RMB 60,000-80,000 band, equivalent to roughly $8,000-11,000 per tonne. That collapse alone removed well over $30/kWh from cell costs across chemistries. Graphite has been stickier, with Chinese export controls on certain anode grades adding a premium for buyers outside Asia. The upshot is that 2026 pricing is far more sensitive to anode and electrolyte supply than to lithium itself.

From cell to module: BMS, thermal hardware and the enclosure

The battery management system is the most misunderstood line item in the stack. A passive BMS on a commodity module costs $2-3/kWh. A high-voltage master architecture with redundant cell monitoring, ISO 26262-style functional safety, active balancing, state-of-health estimation and full SCADA integration can cost $8-12/kWh. For a 5 MWh container, that difference is $30,000-60,000 of hardware, and it is frequently the item where aggressive quotations are shaved.
Thermal management is the second variable. Air cooling costs $2-4/kWh and remains common in residential and small commercial packs. Liquid cooling, now standard for utility-scale containers, runs $5-9/kWh installed once pumps, manifolds, coolant, leak detection and the HVAC plant are counted. Liquid cooling is not optional at high C-rates — an LFP cell cycled at 1C continuously in a 40°C ambient will degrade measurably faster without it — but it is routinely under-specified in low-cost quotations. Buyers comparing numbers should confirm whether HVAC and fire-suppression systems are inside or outside the stated $/kWh.
Enclosures, racks and structural steel add $4-8/kWh depending on IP rating, seismic requirements and whether the product is a cabinet, a container or a rack for a DC block inside a building. Assembly and testing — module stacking, laser welding of busbars, end-of-line hipot and capacity grading — contributes $3-5/kWh in China and 1.5-2.5× that in Europe or North America. Freight, insurance, warranty reserve and supplier margin close out the stack at a further $6-10/kWh for export sales.

A line-by-line cost breakdown for LFP packs

The table below models a Chinese-manufactured LFP DC pack selling at approximately $85/kWh, with indicative US-landed equivalents before tariffs. Percentages vary with format, scale and order size; treat them as a structural map rather than a quotation.
| Cost line item | Share of pack price | China ($/kWh) | US-landed ($/kWh) |
|---|---|---|---|
| Cells (LFP prismatic, 280-320 Ah) | 65-72% | 55-62 | 78-95 |
| Module hardware, busbars, sensing | 3-5% | 3-4 | 5-7 |
| BMS (master + slave, HV architecture) | 5-8% | 4-7 | 10-16 |
| Thermal management (liquid, incl. HVAC) | 4-7% | 3-6 | 7-12 |
| Enclosure, rack, structural, IP rating | 5-8% | 4-7 | 9-15 |
| Contactors, fuses, pre-charge, connectors | 3-5% | 3-4 | 5-8 |
| Assembly, formation, EOL test, yield loss | 3-5% | 3-4 | 6-10 |
| Freight, warranty reserve, margin | 6-10% | 5-9 | 20-35 |
| Delivered DC pack total | 100% | ~85 | ~150-200 |
Two things stand out. First, the US-landed column is not simply China plus freight: the multipliers on BMS, enclosures and assembly reflect genuine differences in labour, engineering and compliance cost, and the margin and warranty lines absorb the working capital and performance guarantees that Western offtakers demand. Second, the AC-system premium sits on top of this — PCS, transformer, MV switchgear and controls add $50-90/kWh in most markets, which is why a turnkey four-hour US BESS regularly prices at $220-280/kWh while the DC block inside it is quoted far lower.

Chemistry choices change the cost per cycle, not just the cost per kWh

LFP has taken roughly half of global EV battery demand and the overwhelming majority of new stationary storage, according to the IEA's 2025 market assessment, on the strength of three numbers: 6,000-8,000 cycles to 80% state of health, materially lower fire risk, and no nickel or cobalt exposure. Its weakness is energy density — 140-160 Wh/kg at pack level versus 200-250 Wh/kg for NMC — which matters for vehicles and hardly at all for a container sitting on concrete.
Comparing LFP and NMC at the pack level therefore misleads. A $95/kWh LFP pack cycling 7,000 times delivers energy at roughly $0.0136/kWh per cycle of throughput; a $120/kWh NMC pack cycling 3,500 times delivers at roughly $0.0343/kWh per cycle, before accounting for augmentation. The LFP system wins by a factor of 2.5 on throughput economics, which is why the storage market has largely converged on it. The trade-off appears in cold climates, where LFP charge acceptance at sub-zero temperatures requires active heating and erodes both round-trip efficiency and usable capacity.
Practical AC round-trip efficiency for a well-integrated LFP system lands at 85-89% including PCS, transformer and auxiliary loads; 88-92% DC-to-DC at the pack boundary is typical for new cells. Buyers who see round-trip efficiency quoted above 92% for a full AC system should ask whether parasitic cooling loads have been included. Efficiency is a cost line disguised as a specification: two percentage points of round-trip loss on a 100 MWh annual throughput is worth more than the price difference between most competing bids.

Raw materials and the 2026 policy shock

Two exogenous factors will shape 2026 pricing more than manufacturing improvements. The first is US trade policy: the Section 301 tariff on Chinese non-EV lithium-ion batteries rises from 7.5% to 25% on 1 January 2026, following the EV battery increase that took effect in 2024. For a DC block imported at $100/kWh, that step adds roughly $17.50/kWh at the border, and it lands on a supply chain that still sources a large share of cells, graphite and electrolyte from China even when final assembly is domestic.
The second is the manufacturing credit structure. Under Section 45X, US-produced battery cells earn $35/kWh and modules $10/kWh, with a $45/kWh pathway for cells sold without a module. That credit is large enough to make domestic cell production viable at a cost structure that would otherwise be uncompetitive with Chinese imports — but it phases down and carries FEOC rules that disqualify material sourced from defined foreign entities of concern. Developers claiming the investment tax credit must trace their supply chain accordingly, and many 2026 procurement decisions hinge on documentation rather than price.
The practical effect is a two-tier global market. China's domestic DC block pricing has compressed toward $50-60/kWh in competitive utility tenders, while US-landed DC pricing holds well above $150/kWh and European pricing sits between the two. BloombergNEF's outlook still points to roughly $80/kWh globally by 2030, but that trajectory assumes tariff regimes do not fracture further — an assumption no serious procurement team is currently making.

What buyers should actually interrogate in a quotation

The first question is scope. A quoted $/kWh may refer to nameplate DC energy, usable AC energy after depth-of-discharge limits, or nameplate at beginning of life. An LFP system quoted at $180/kWh on nameplate DC with 90% usable and 95% end-of-first-year capacity is effectively $210/kWh on delivered AC energy. Normalising every bid to usable AC energy at year one, and again at year ten, collapses most of the apparent spread between vendors.
The second is warranty structure. A ten-year warranty is close to meaningless without a throughput cap. Ask for the guaranteed MWh throughput, the end-of-warranty capacity retention (typically 70% for utility, 80% for commercial), the augmentation obligations, and who pays for labour and logistics on a replacement. Warranty reserve is 2-4% of pack price in a well-run vendor's model; a bid with no visible reserve is either subsidised by a balance sheet or under-provisioned.
The third is thermal and safety documentation. For utility-scale projects, UL 9540A test reports, cell-level UN 38.3 transport certification and IEC 62619 compliance should be available before contract, not after. For C&I systems, confirm whether the PCS is integrated or separate — the interaction between battery and inverter affects efficiency, response time and warranty boundaries, and it is worth understanding how the two are matched in any inverter and storage architecture under consideration.
Finally, ask about supply chain concentration. Cell supplier, cathode origin, separator film source and BMS silicon vendor should all be disclosed. A pack is only as bankable as its least substitutable component, and the ability to swap a single-source BMS board without re-certifying the whole container is worth real money over a fifteen-year asset life. Project references remain the fastest sanity check; installed system case studies reveal more about real degradation and service response than any datasheet.

Where the cost curve goes next

Cell-level costs in China are already approaching the floor set by materials and energy, with leading LFP producers reporting cell prices in the $45-55/kWh range for large orders. The remaining opportunities sit in the balance of pack: higher-density 314 Ah and 587 Ah cells that cut enclosure and busbar content per kWh, DC-side architectures that eliminate one stage of conversion, and container-level integration that reduces field labour. Each is worth $3-8/kWh, not $30.
The bigger swing factor is manufacturing scale outside China. Every gigafactory that reaches stable 95%+ yield removes a cost penalty currently embedded in Western pack prices. That process is slower than the announcement pipeline suggests, but it is happening: US cell capacity online has roughly quadrupled since 2022. For buyers, the strategic implication is straightforward — the chemistry is settled, the supply chain is not, and the difference between a $108/kWh pack and a $200/kWh pack is now decided by tariffs, logistics, certification and yield rather than by electrochemistry.