Battery chemistry discussions usually collapse into a single axis, energy density, and then stop. That axis explains why nickel-rich cells went into premium vehicles. It does not explain why almost every grid battery ordered in the last three years is LFP, or why sodium-ion started shipping into markets where lithium was never the bottleneck.
Naming convention first, because it causes confusion. Commercial chemistries are named after the cathode. The anode is graphite in the overwhelming majority of cells, sometimes blended with silicon. When someone says "an NMC cell", they are describing one electrode and leaving the other implied.
The comparison
Figures below are cell level for commercial 2026 production, not pack level and not laboratory records. Pack-level energy typically lands 10 to 35 per cent below cell level: cell-to-pack LFP designs sit at the low end of that range, module-based nickel-rich packs at the high end.
| LFP | LMFP | NMC (622 to 811) | NCA | Sodium-ion | |
|---|---|---|---|---|---|
| Cathode | Lithium iron phosphate | Lithium manganese iron phosphate | Lithium nickel manganese cobalt oxide | Lithium nickel cobalt aluminium oxide | Layered oxide, polyanion or Prussian white |
| Nominal voltage | 3.2 to 3.3 V | 3.7 to 3.9 V | 3.6 to 3.7 V | 3.6 V | 3.0 to 3.2 V |
| Cell specific energy | 90 to 205 Wh/kg | 180 to 230 Wh/kg | 200 to 300 Wh/kg | 240 to 300 Wh/kg | 100 to 175 Wh/kg |
| Typical cycle life | 3,000 to 8,000+ | 2,000 to 4,000 | 1,000 to 3,000 | 800 to 2,000 | 3,000 to 6,000 claimed |
| Thermal decomposition onset | Highest of the group | High | Lower, falls as nickel rises | Lowest of the group | Comparable to LFP |
| Critical materials | None of Ni, Co | None of Ni, Co | Nickel, cobalt | Nickel, cobalt | None of Li, Ni, Co |
| Voltage curve | Very flat | Flat with a step | Sloped | Sloped | Sloped |
| Low-temperature performance | Weak | Weak to moderate | Moderate | Moderate | Strong |
| Where it wins | Stationary storage, mass-market EV, buses | Mid-range EV, seeking LFP cost with more range | Premium EV, tools, aviation | Premium EV, high specific energy | Stationary, entry mobility, cold climates, start-stop |
The upper ends of the specific energy rows track the best commercial cells: the IEA's Global EV Outlook 2026 puts top LFP cells at 205 Wh/kg and top sodium-ion cells at 175 Wh/kg. Both figures move every product generation.
Why LFP took stationary storage
LFP is no longer the challenger chemistry in any segment. The IEA's Global EV Outlook 2026 reports that LFP passed 55 per cent of global EV battery deployment in 2025, up from about 50 per cent in 2024, and in stationary storage its dominance is close to total. Three reasons explain the takeover, and only one of them is cost.
The olivine phosphate structure is thermally more stable than layered oxides. Oxygen release during thermal decomposition is the mechanism that turns a single cell failure into a propagating event, and phosphate releases far less of it at a higher onset temperature. For a container holding several megawatt hours next to a substation, that margin is worth more than 40 Wh/kg.
Cycle life is high and predictable, which suits an asset whose revenue model is throughput. A storage developer signing a twenty-year offtake needs a warranty curve, not a peak number.
Then cost. Removing nickel and cobalt removes both price and price volatility, and it removes the supply chain due diligence burden that comes with cobalt.
The trade-off shows up in the BMS, not the datasheet. LFP's open-circuit voltage curve is nearly flat across the middle of the range, so voltage tells you very little about state of charge. Estimation leans on coulomb counting, which drifts, so LFP systems need periodic full charges to re-establish a reference. Anyone specifying LFP for an application that never reaches full charge should raise that with the integrator early.
Where nickel-rich still wins
Mass and volume. An aircraft, a performance vehicle or a device with a fixed enclosure pays real money for every kilogram and litre, and 280 Wh/kg against 170 Wh/kg is not a marginal difference.
The cost of that density is a narrower operating window. Upper cut-off voltage matters more, high state of charge dwell matters more, and thermal runaway onset is lower, which pushes cost back into pack engineering: cooling, propagation barriers, venting design and detection.
Nickel content is the dial. Moving from NMC 622 to 811 raises specific energy and cuts cobalt, and simultaneously reduces thermal stability and increases surface reactivity. Manufacturers manage that with coatings, single-crystal morphologies and electrolyte additives, all of which add process cost.
LMFP and the middle ground
LMFP substitutes manganese into the iron phosphate lattice to lift the voltage plateau from roughly 3.4 V to the region of 4.1 V, which raises energy without adding cobalt. It keeps much of the thermal advantage of the phosphate family.
The engineering problems are manganese dissolution, lower electronic conductivity and a two-plateau voltage curve that complicates state of charge estimation further. Products entering the market often blend LMFP with NMC to smooth the curve and lift power. Treat supplier energy claims as a target rather than a delivered figure until you see cycle data at temperature.
Sodium-ion in one paragraph of honesty
Sodium is abundant and cheap, and sodium-ion cells can use aluminium current collectors on both electrodes rather than copper on the anode side, which cuts cost and mass. Sodium-ion also holds capacity at low temperature far better than lithium iron phosphate, which matters for northern grid storage and cold-climate vehicles, and it tolerates discharge to zero volts for transport. Specific energy sits below LFP for now. The realistic near-term positions are stationary storage, two and three-wheelers, entry-level city vehicles and twelve-volt automotive systems. Claims about cycle life are mostly supplier figures rather than independently published, so treat them accordingly.
The anode side, briefly
Graphite remains the default. Silicon blends of a few per cent are common, silicon-dominant anodes are shipping in niche high-energy applications, and lithium metal is the target for most solid-state programmes. Every step along that path buys specific energy and pays with volume expansion, first-cycle efficiency loss and calendar life, which is why silicon content has risen slowly rather than in a jump.
How to choose, in practice
Start from the binding constraint rather than the chemistry.
If the constraint is mass or volume, you are in nickel-rich territory and your engineering budget goes into thermal design. If the constraint is cost per delivered kWh over twenty years, you are in phosphate territory and your budget goes into state of charge management and BMS calibration. If the constraint is cold ambient temperature or lithium supply exposure, sodium-ion deserves a real evaluation rather than a footnote. If the constraint is peak power rather than energy, cell format, tab design and internal resistance will matter more than which cathode you picked.
One regulatory factor now sits alongside the technical ones. The EU Battery Regulation introduces carbon footprint declaration and, on a staged timetable, recycled content requirements for cobalt, lead, lithium and nickel. That changes the total cost comparison by chemistry in a way that was not present in earlier procurement cycles, and it rewards chemistries with shorter and less carbon-intensive upstream chains.
Informational and educational content only. Not professional, financial, legal, or engineering advice.